From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id EBEF94C6E for ; Wed, 26 Mar 2025 00:05:10 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947518; cv=none; b=hZW3WTvycK/mHHHtkAAvcFD+1OwQGGbeQCRt4NlbL4zoslz7UYpxnlMqJd3g52adGPShsa7DAlgUMcBTRDohS2oWZC0iH9+0zXU7b6WYS/J/2kYkVkWx6I3rJbQup0NIxgpSq3Dl9RDIc5FOlAX6rd2XLBtNbaAIdBsRdyGmNus= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947518; c=relaxed/simple; bh=4FPUsIkI2Jj+5+9O1ngomErPVk5nYs3hF77c/cFAskI=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=UUqWUHmR5OazAOaGbhtkrDIVX6ZBA+diHyL1T2QEdMhbM4QpHCKbBdu7u3bSloliS6Z/EeKD5jdn4W/GEqNRsxKoW4jgD8p8rZiI3F9v+2A01vnE5TSV/+fxRzl82NWbeOs4+hJxPKv9w62d78pcuPdHMmNcLO+OC8JPYB2eNdk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=ZLcDYosQ; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="ZLcDYosQ" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff798e8c93so10555203a91.2 for ; Tue, 25 Mar 2025 17:05:10 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947510; x=1743552310; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=eF9J9QMtnwfKnqx+D/oCj2ZUCf9Pw341GFP+/jL0cJs=; b=ZLcDYosQeNono/puPRk0RnvzCOV55BNJd99YQJW0KFM25/iR07u3Pn34xU/5HSaAJE 4bOwVshtgBAaTHQ0izutq96VF1lnx9NgiLCH/0dzVjo3KJ+3u0lQ3HgvIYcq3ZnkBPLO W2shZWr7lvLAa4J1febgyXLyu/o86dWDsKEqJPv6I0/THdfFNj4/xykld97OUvwTIsjS uwUReFz5qA93Pgx6ZqLdRhCQXeBmy0rn12Nhu0u18kwQ2BYaXpvqf+rKYYtp76AiBFFE WalbIirFt/hMgM5urKwXg7NpwbJOCTKfg0XLR+r0f0obTJ+0KEM6MnBPIcuUH0Pd4pl/ EvRQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947510; x=1743552310; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=eF9J9QMtnwfKnqx+D/oCj2ZUCf9Pw341GFP+/jL0cJs=; b=FZ6s3t4CwoiXJKJBnV9I2gXwCfcaaD3ktXeCbM282GdjwNDPePg1bQfdb8jciet2gp 4qDsZ21ezXrQkry1gplNm5nLLUssuLEVYLmjm0d1QXOCdl6TlL9rl2a5B1Ro5PxoHK+m nxVfeS4lp7LSCTRiUttPeTArE/fFSsZx7cF79lJjkFsI2C4k+5yp8UOuvxTFv1vpOq/E /90OcFjrryJc8lhmOh3tkhvBIKBQO3tP3uePoxeKAEXp2kzypDB3GK6d/d9djASNwT+o TFvQPjoGFH3G9+K8JiIpngsvaMbYkrGqvCcT1ao66X6F2nbgFTLXpp+wLLtsLshCadZZ 6apA== X-Forwarded-Encrypted: i=1; AJvYcCXrRK/i5J143Xj67vqEFuy+SnHW2pp7l3WVozgkvR7CnO5pelEj6z/oYvLt734fOBfw9i4caNBrabwKb20=@vger.kernel.org X-Gm-Message-State: AOJu0Yz+H0KRHq18bvBapg5PB8N403MdPkSFtXzhDHPz8/gI1CEEpkWx NUpJ9HTign3mcSqhA2762wsfKMCCTbTD6E5XmFzovaVY4yPu4yxmr4wf9jYy9Eykg0Hw2bGJGSh umIge4g== X-Google-Smtp-Source: AGHT+IFCvtOqW1Sj003oREnKSMgtUZhZ4R5zD2OQh+tOI4rd7QM8pUls+7dLANxa2lyLqzLdsALr0O88JjIZ X-Received: from pjbmf11.prod.google.com ([2002:a17:90b:184b:b0:2ef:d283:5089]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:554f:b0:2ee:c6c8:d89f with SMTP id 98e67ed59e1d1-3030fe94313mr35309773a91.14.1742947510190; Tue, 25 Mar 2025 17:05:10 -0700 (PDT) Date: Tue, 25 Mar 2025 17:03:59 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-2-irogers@google.com> Subject: [PATCH v3 01/35] perf vendor events: Update alderlake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.28 to v1.29. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/alderlake/adl-metrics.json | 485 +++++++++--------- .../pmu-events/arch/x86/alderlake/cache.json | 77 +++ .../pmu-events/arch/x86/alderlake/memory.json | 55 ++ .../pmu-events/arch/x86/alderlake/other.json | 196 ------- .../arch/x86/alderlake/pipeline.json | 67 ++- tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- 6 files changed, 441 insertions(+), 441 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/alderlake/adl-metrics.json b/to= ols/perf/pmu-events/arch/x86/alderlake/adl-metrics.json index 147379cae37b..2b88590e3756 100644 --- a/tools/perf/pmu-events/arch/x86/alderlake/adl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/alderlake/adl-metrics.json @@ -103,7 +103,7 @@ "MetricExpr": "tma_core_bound", "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", - "MetricThreshold": "(tma_allocation_restriction >0.10) & ((tma_cor= e_bound >0.10) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -113,7 +113,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALL@ / (5 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound >0.10)", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%", @@ -125,7 +125,7 @@ "MetricExpr": "(5 * cpu_atom@CPU_CLK_UNHALTED.CORE@ - (cpu_atom@TO= PDOWN_FE_BOUND.ALL@ + cpu_atom@TOPDOWN_BE_BOUND.ALL@ + cpu_atom@TOPDOWN_RET= IRING.ALL@)) / (5 * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation >0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%", @@ -136,7 +136,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_DETECT@ / (5 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "MetricThreshold": "(tma_branch_detect >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -146,7 +146,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MISPREDICT@ / (5 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts >0.05) & ((tma_bad_spe= culation >0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -156,7 +156,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_RESTEER@ / (5 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", - "MetricThreshold": "(tma_branch_resteer >0.05) & ((tma_ifetch_late= ncy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -165,7 +165,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.CISC@ / (5 * cpu_atom@CPU= _CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc >0.05) & ((tma_ifetch_bandwidth >0.1= 0) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -174,7 +174,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS@ / (5 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound >0.10) & ((tma_backend_bound >= 0.10))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -184,7 +184,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.DECODE@ / (5 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", - "MetricThreshold": "(tma_decode >0.05) & ((tma_ifetch_bandwidth >0= .10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -193,7 +193,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.FASTNUKE@ / (5 * c= pu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", - "MetricThreshold": "(tma_fast_nuke >0.05) & ((tma_machine_clears >= 0.05) & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -203,7 +203,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ALL@ / (5 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound >0.20)", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -213,7 +213,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ICACHE@ / (5 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -222,7 +222,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH@ / (5 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", - "MetricThreshold": "(tma_ifetch_bandwidth >0.10) & ((tma_frontend_= bound >0.20))", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -232,7 +232,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_LATENCY@ / (5 * = cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", - "MetricThreshold": "(tma_ifetch_latency >0.15) & ((tma_frontend_bo= und >0.20))", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -567,7 +567,7 @@ "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "cpu_atom@CPU_CLK_UNHALTED.CORE_P@ / cpu_atom@CPU_CL= K_UNHALTED.CORE@", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9", "Unit": "cpu_atom" }, { @@ -606,7 +606,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ITLB@ / (5 * cpu_atom@CPU= _CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses >0.05) & ((tma_ifetch_latency= >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -615,7 +615,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS@ / = (5 * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears >0.05) & ((tma_bad_specula= tion >0.15))", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -625,7 +625,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.MEM_SCHEDULER@ / (5 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", - "MetricThreshold": "(tma_mem_scheduler >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -634,7 +634,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER@ / (5 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", - "MetricThreshold": "(tma_non_mem_scheduler >0.10) & ((tma_resource= _bound >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -643,7 +643,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.NUKE@ / (5 * cpu_a= tom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", - "MetricThreshold": "(tma_nuke >0.05) & ((tma_machine_clears >0.05)= & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -652,7 +652,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.OTHER@ / (5 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", - "MetricThreshold": "(tma_other_fb >0.05) & ((tma_ifetch_bandwidth = >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -661,7 +661,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.PREDECODE@ / (5 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", - "MetricThreshold": "(tma_predecode >0.05) & ((tma_ifetch_bandwidth= >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -670,7 +670,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REGISTER@ / (5 * cpu_atom= @CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", - "MetricThreshold": "(tma_register >0.10) & ((tma_resource_bound >0= .20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -679,7 +679,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REORDER_BUFFER@ / (5 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", - "MetricThreshold": "(tma_reorder_buffer >0.10) & ((tma_resource_bo= und >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -688,7 +688,7 @@ "MetricExpr": "tma_backend_bound - tma_core_bound", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", - "MetricThreshold": "(tma_resource_bound >0.20) & ((tma_backend_bou= nd >0.10))", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -699,7 +699,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_RETIRING.ALL@ / (5 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "(tma_retiring >0.75)", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -709,7 +709,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.SERIALIZATION@ / (5 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", - "MetricThreshold": "(tma_serialization >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -721,7 +721,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(cpu_core@UOPS_DISPATCHED.PORT_0@ + cpu_core@UOPS_D= ISPATCHED.PORT_1@ + cpu_core@UOPS_DISPATCHED.PORT_5_11@ + cpu_core@UOPS_DIS= PATCHED.PORT_6@) / (5 * tma_info_core_core_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -734,13 +734,13 @@ "MetricExpr": "78 * cpu_core@ASSISTS.ANY@ / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_threa= d_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -751,7 +751,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-be\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -768,13 +768,13 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20", @@ -791,7 +791,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -800,7 +800,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_= bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_d= ram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_st= ore_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_load= s + tma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_= l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_s= tore_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sha= ring + tma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_mem= ory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound = + tma_dram_bound + tma_store_bound)) * (tma_store_latency / (tma_store_late= ncy + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtl= b_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependen= cy + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_= bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma= _l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_fb= _full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tm= a_store_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tm= a_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_= latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_s= tore_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharin= g + tma_split_stores + tma_store_latency + tma_streaming_stores)) + tma_mem= ory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store= + tma_false_sharing + tma_split_stores + tma_store_latency + tma_streaming= _stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -809,16 +809,16 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy", + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: ", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_= clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_br= anch_mispredicts) / (tma_mispredicts_resteers + tma_clears_resteers + tma_u= nknown_branches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_reste= ers + tma_ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth *= tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_co= de", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cl= ears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_bran= ch_mispredicts) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unk= nown_branches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_miss= es + tma_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * t= ma_ms / (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code= ", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", @@ -826,7 +826,7 @@ }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_s= witches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_rest= eers * tma_other_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_r= esteers + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses= + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_= dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ls= d + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other= _nukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cp= u_core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / = (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_mic= rocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * = (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_restee= rs * tma_other_mispredicts / tma_branch_mispredicts) / (tma_clears_resteers= + tma_mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers= + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_m= s_switches) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite = + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_bra= nch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_n= ukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cpu_= core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / (t= ma_divider + tma_ports_utilization + tma_serializing_operation) + tma_micro= code_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * (t= ma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -835,7 +835,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound += tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_store_latency + tma_false_sharing + tma_split_stores + tma_= streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dram_bound= + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tma_sto= re_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -844,16 +844,16 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears", + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -866,11 +866,11 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls", + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls.", "Unit": "cpu_core" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (cpu_core@BR_INST_RETIRED.ALL= _BRANCHES@ + 2 * cpu_core@BR_INST_RETIRED.NEAR_CALL@ + cpu_core@INST_RETIRE= D.NOP@) / tma_info_thread_slots - tma_microcode_sequencer / (tma_few_uops_i= nstructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_seque= ncer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -879,7 +879,7 @@ }, { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "cpu_core@topdown\\-br\\-mispredict@ / (cpu_core@top= down\\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-re= tiring@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -893,26 +893,26 @@ "MetricExpr": "cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_= thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C01@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C02@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -921,7 +921,7 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRE= D.MS_FLOWS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -931,26 +931,26 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_D= EMAND_CODE_RD@ / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -959,7 +959,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -968,35 +968,35 @@ "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * cpu_core@ITLB_MISSES.WALK_COMP= LETED_2M_4M@ / (cpu_core@ITLB_MISSES.WALK_COMPLETED_4K@ + cpu_core@ITLB_MIS= SES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * cpu_core@ITLB_MISSES.WALK_COMP= LETED_4K@ / (cpu_core@ITLB_MISSES.WALK_COMPLETED_4K@ + cpu_core@ITLB_MISSES= .WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((28 * tma_info_system_core_frequency - 3 * tma_inf= o_system_core_frequency) * (cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (c= pu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DEMAND_DATA_R= D.L3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FW= D@))) + (27 * tma_info_system_core_frequency - 3 * tma_info_system_core_fre= quency) * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@) * (1 + cpu_core@MEM_= LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_info_t= hread_clks", + "MetricExpr": "(25 * tma_info_system_core_frequency * (cpu_core@ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP= _HITM@ / (cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEM= AND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD@))) + 24 * tma_info_system_core_frequ= ency * cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@) * (1 + cpu_core@MEM_LOA= D_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_info_thre= ad_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1007,26 +1007,26 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "(27 * tma_info_system_core_frequency - 3 * tma_info= _system_core_frequency) * (cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ + = cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (1 - cpu_core@OCR.DEMAND_DATA_= RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ + c= pu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD@))) * (1 + cpu_core@ME= M_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_info= _thread_clks", + "MetricExpr": "24 * tma_info_system_core_frequency * (cpu_core@MEM= _LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_F= WD@ * (1 - cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP= _HIT_WITH_FWD@))) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_L= OAD_RETIRED.L1_MISS@ / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - = cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks = / 2", + "MetricExpr": "(cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cp= u_core@INST_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1036,7 +1036,7 @@ "MetricExpr": "cpu_core@ARITH.DIV_ACTIVE@ / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1046,7 +1046,7 @@ "MetricExpr": "cpu_core@MEMORY_ACTIVITY.STALLS_L3_MISS@ / tma_info= _thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1057,7 +1057,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1066,28 +1066,28 @@ "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / tma_in= fo_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu_core@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\= \=3D0x1@ + cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@, max(cpu_core@CYCLE_ACTIV= ITY.CYCLES_MEM_ANY@ - cpu_core@MEMORY_ACTIVITY.CYCLES_L1D_MISS@, 0)) / tma_= info_thread_clks", + "MetricExpr": "min(7 * cpu_core@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\= \=3D1@ + cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@, max(cpu_core@CYCLE_ACTIVIT= Y.CYCLES_MEM_ANY@ - cpu_core@MEMORY_ACTIVITY.CYCLES_L1D_MISS@, 0)) / tma_in= fo_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu_core@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\= =3D0x1@ + cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@) / tma_info_core_core_clk= s", + "MetricExpr": "(7 * cpu_core@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\= =3D1@ + cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1096,8 +1096,8 @@ "MetricExpr": "28 * tma_info_system_core_frequency * cpu_core@OCR.= DEMAND_RFO.L3_HIT.SNOOP_HITM@ / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1118,18 +1118,18 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = CPU was stalled due to Frontend latency issues", - "MetricExpr": "topdown\\-fetch\\-lat / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) - cpu_core@I= NT_MISC.UOP_DROPPING@ / tma_info_thread_slots", + "MetricExpr": "cpu_core@topdown\\-fetch\\-lat@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) - cpu_core@INT_MISC.UOP_DROPPING@ / t= ma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend= _bound_group", "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1149,7 +1149,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1159,16 +1159,16 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "cpu_core@ARITH.FPDIV_ACTIVE@ / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1177,8 +1177,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.SCALAR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma= _port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1187,8 +1187,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1197,8 +1197,8 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1207,41 +1207,41 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots wh= ere the processor's Frontend undersupplies its Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-fe\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) - cpu_core@IN= T_MISC.UOP_DROPPING@ / tma_info_thread_slots", + "MetricExpr": "cpu_core@topdown\\-fe\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) - cpu_core@INT_MISC.UOP_DROPPING@ / tm= a_info_thread_slots", "MetricGroup": "BvFB;BvIO;Default;PGO;TmaL1;TopdownL1;tma_L1_group= ", "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.MACRO_= FUSED@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "cpu_core@topdown\\-heavy\\-ops@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1250,8 +1250,8 @@ "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / tma_info_thread_clks= ", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1264,7 +1264,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_NTAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", @@ -1272,7 +1272,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", @@ -1280,15 +1280,15 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.INDIRECT@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.RET@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -1320,7 +1320,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -1329,7 +1329,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -1338,10 +1338,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: ", "Unit": "cpu_core" }, { @@ -1412,12 +1413,12 @@ "MetricExpr": "(cpu_core@FP_ARITH_DISPATCHED.PORT_0@ + cpu_core@FP= _ARITH_DISPATCHED.PORT_1@ + cpu_core@FP_ARITH_DISPATCHED.PORT_5@) / (2 * tm= a_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)", + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n).", "Unit": "cpu_core" }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp", "Unit": "cpu_core" @@ -1432,22 +1433,22 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost", "Unit": "cpu_core" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D1@", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc", "Unit": "cpu_core" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / cpu_core@ICACHE_DATA= .STALLS\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / cpu_core@ICACHE_DATA= .STALLS\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency", "Unit": "cpu_core" @@ -1497,14 +1498,14 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node", + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node.", "Unit": "cpu_core" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "cpu_core@BR_INST_RETIRED.ALL_BRANCHES@ / cpu_core@B= R_INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch", @@ -1524,7 +1525,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW", + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW.", "Unit": "cpu_core" }, { @@ -1533,7 +1534,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1542,7 +1543,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1551,7 +1552,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1560,7 +1561,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1623,7 +1624,7 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 13", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp", "Unit": "cpu_core" }, @@ -1769,7 +1770,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp", "Unit": "cpu_core" @@ -1849,7 +1850,7 @@ }, { "BriefDescription": "", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / (cpu_core@UOPS_EXE= CUTED.CORE_CYCLES_GE_1@ / 2 if #SMT_on else cpu_core@UOPS_EXECUTED.THREAD\\= ,cmask\\=3D0x1@)", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / (cpu_core@UOPS_EXE= CUTED.CORE_CYCLES_GE_1@ / 2 if #SMT_on else cpu_core@UOPS_EXECUTED.THREAD\\= ,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute", "Unit": "cpu_core" @@ -1880,20 +1881,20 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@ASSISTS.ANY@", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)", "Unit": "cpu_core" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire", "Unit": "cpu_core" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1", @@ -1946,23 +1947,22 @@ }, { "BriefDescription": "Instructions per Far Branch ( Far Branches ap= ply upon transition from application to operating system, handling interrup= ts, exceptions) [lower number means higher occurrence rate]", - "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / BR_INST_RETIRED.FAR_BR= ANCH:u", + "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.FAR_BRANCH@u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000", + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6", "Unit": "cpu_core" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@INS= T_RETIRED.ANY_P@k", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / cpu_core@CPU_CLK_UNHA= LTED.THREAD@", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@CPU= _CLK_UNHALTED.THREAD@", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05", @@ -2030,7 +2030,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD@", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks", @@ -2041,7 +2041,6 @@ "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { @@ -2049,7 +2048,7 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_ISSU= ED.ANY@", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage", + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage.", "Unit": "cpu_core" }, { @@ -2061,14 +2060,14 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "cpu_core@TOPDOWN.SLOTS@", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (cpu_core@TOPDOWN.SLOTS@ /= 2) if #SMT_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization", "Unit": "cpu_core" @@ -2086,15 +2085,15 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@BR_= INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5", + "MetricThreshold": "tma_info_thread_uptb < 9", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2104,7 +2103,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2113,8 +2112,8 @@ "MetricExpr": "(cpu_core@INT_VEC_RETIRED.ADD_128@ + cpu_core@INT_V= EC_RETIRED.VNNI_128@) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_port_0, tma_port_1,= tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2123,8 +2122,8 @@ "MetricExpr": "(cpu_core@INT_VEC_RETIRED.ADD_256@ + cpu_core@INT_V= EC_RETIRED.MUL_256@ + cpu_core@INT_VEC_RETIRED.VNNI_256@) / (tma_retiring *= tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_port_0, tma_por= t_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2133,8 +2132,8 @@ "MetricExpr": "cpu_core@ICACHE_TAG.STALLS@ / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2143,7 +2142,7 @@ "MetricExpr": "max((cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ - cpu_co= re@MEMORY_ACTIVITY.STALLS_L1D_MISS@) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2153,7 +2152,7 @@ "MetricExpr": "min(2 * (cpu_core@MEM_INST_RETIRED.ALL_LOADS@ - cpu= _core@MEM_LOAD_RETIRED.FB_HIT@ - cpu_core@MEM_LOAD_RETIRED.L1_MISS@) * 20 /= 100, max(cpu_core@CYCLE_ACTIVITY.CYCLES_MEM_ANY@ - cpu_core@MEMORY_ACTIVIT= Y.CYCLES_L1D_MISS@, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2163,7 +2162,7 @@ "MetricExpr": "(cpu_core@MEMORY_ACTIVITY.STALLS_L1D_MISS@ - cpu_co= re@MEMORY_ACTIVITY.STALLS_L2_MISS@) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2173,7 +2172,7 @@ "MetricExpr": "3 * tma_info_system_core_frequency * cpu_core@MEM_L= OAD_RETIRED.L2_HIT@ * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM= _LOAD_RETIRED.L1_MISS@ / 2) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2183,18 +2182,18 @@ "MetricExpr": "(cpu_core@MEMORY_ACTIVITY.STALLS_L2_MISS@ - cpu_cor= e@MEMORY_ACTIVITY.STALLS_L3_MISS@) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(12 * tma_info_system_core_frequency - 3 * tma_info= _system_core_frequency) * (cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2)) / tma= _info_thread_clks", + "MetricExpr": "9 * tma_info_system_core_frequency * (cpu_core@MEM_= LOAD_RETIRED.L3_HIT@ * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@ME= M_LOAD_RETIRED.L1_MISS@ / 2)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2203,19 +2202,19 @@ "MetricExpr": "cpu_core@DECODE.LCP@ / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2234,7 +2233,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2243,34 +2242,34 @@ "MetricExpr": "cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@ / tma_info_t= hread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_1G@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_2M_4M@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_C= OMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_4K@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2279,7 +2278,7 @@ "MetricExpr": "(16 * max(0, cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ = - cpu_core@L2_RQSTS.ALL_RFO@) + cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu= _core@MEM_INST_RETIRED.ALL_STORES@ * (10 * cpu_core@L2_RQSTS.RFO_HIT@ + min= (cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFFCORE_REQUESTS_OUTSTANDING.C= YCLES_WITH_DEMAND_RFO@))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2290,7 +2289,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2301,16 +2300,16 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_c= lks", + "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clk= s", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2320,34 +2319,34 @@ "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD@) / tma_info_thread_clks - tm= a_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-mem\\-bound@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * cpu_core@MISC2_RETIRED.LFENCE@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * cpu_core@MEM_UOP_RETIRED.ANY= @ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -2370,7 +2369,7 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * cpu_= core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2386,18 +2385,18 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "max(cpu_core@IDQ.MS_CYCLES_ANY@, cpu_core@UOPS_RETI= RED.MS\\,cmask\\=3D0x1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISS= UED.ANY@)) / tma_info_core_core_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "max(cpu_core@IDQ.MS_CYCLES_ANY@, cpu_core@UOPS_RETI= RED.MS\\,cmask\\=3D1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUE= D.ANY@)) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -2406,10 +2405,10 @@ }, { "BriefDescription": "This metric estimates the fraction of cycles = when the CPU was stalled due to switches of uop delivery to the Microcode S= equencer (MS)", - "MetricExpr": "3 * cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1\\,edge= \\=3D0x1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUED.ANY@) / tm= a_info_thread_clks", + "MetricExpr": "3 * cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1\\,edge@ = / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUED.ANY@) / tma_info_thr= ead_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tm= a_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_mac= hine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_o= peration", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2420,7 +2419,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2429,7 +2428,7 @@ "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.NOP@ /= (tma_retiring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2445,20 +2444,20 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - cpu_core@BR_MISP_= RETIRED.ALL_BRANCHES@ / (cpu_core@INT_MISC.CLEARS_COUNT@ - cpu_core@MACHINE= _CLEARS.COUNT@)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - cpu_core@MACHINE_CLEA= RS.MEMORY_ORDERING@ / cpu_core@MACHINE_CLEARS.COUNT@), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2468,7 +2467,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2478,7 +2477,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_12= 8b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_5, tma_po= rt_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2488,7 +2487,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_= 0, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2498,7 +2497,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128= b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= t_5, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2507,8 +2506,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (cp= u_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVITY.2_= 3_PORTS_UTIL@)) / tma_info_thread_clks if cpu_core@ARITH.DIV_ACTIVE@ < cpu_= core@CYCLE_ACTIVITY.STALLS_TOTAL@ - cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ e= lse (cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTI= VITY.2_3_PORTS_UTIL@) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2517,8 +2516,8 @@ "MetricExpr": "(cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ + max(cpu= _core@RS.EMPTY_RESOURCE@ - cpu_core@RESOURCE_STALLS.SCOREBOARD@, 0)) / tma_= info_thread_clks * (cpu_core@CYCLE_ACTIVITY.STALLS_TOTAL@ - cpu_core@EXE_AC= TIVITY.BOUND_ON_LOADS@) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2527,7 +2526,7 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2538,8 +2537,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.2_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma= _int_vector_256b, tma_port_0, tma_port_1, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2549,7 +2548,7 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.CYCLES_GE_3@ / tma_info_thre= ad_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2557,7 +2556,7 @@ { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-retiring@ / (cpu_core@topdown\\-= fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiring@= + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2571,7 +2570,7 @@ "MetricExpr": "cpu_core@RESOURCE_STALLS.SCOREBOARD@ / tma_info_thr= ead_clks + tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2581,8 +2580,8 @@ "MetricExpr": "tma_light_operations * cpu_core@INT_VEC_RETIRED.SHU= FFLES@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2592,7 +2591,7 @@ "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.PAUSE@ / tma_info_thread_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2603,7 +2602,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2612,8 +2611,8 @@ "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ / tma_info_= core_core_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2622,7 +2621,7 @@ "MetricExpr": "(cpu_core@XQ.FULL_CYCLES@ + cpu_core@L1D_PEND_MISS.= L2_STALLS@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2632,8 +2631,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.BOUND_ON_STORES@ / tma_info_t= hread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2642,8 +2641,8 @@ "MetricExpr": "13 * cpu_core@LD_BLOCKS.STORE_FORWARD@ / tma_info_t= hread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2652,8 +2651,8 @@ "MetricExpr": "(cpu_core@MEM_STORE_RETIRED.L2_HIT@ * 10 * (1 - cpu= _core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.ALL_STORES@)= + (1 - cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.A= LL_STORES@) * min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFFCORE_REQUE= STS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO@)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2672,7 +2671,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2681,34 +2680,34 @@ "MetricExpr": "cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@ / tma_info_= core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_1G@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_2M_4M@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_= core@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.W= ALK_COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_4K@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2717,7 +2716,7 @@ "MetricExpr": "9 * cpu_core@OCR.STREAMING_WR.ANY_RESPONSE@ / tma_i= nfo_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2727,7 +2726,7 @@ "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / tma_info= _thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2737,8 +2736,8 @@ "MetricExpr": "tma_retiring * cpu_core@UOPS_EXECUTED.X87@ / cpu_co= re@UOPS_EXECUTED.THREAD@", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%", "Unit": "cpu_core" } diff --git a/tools/perf/pmu-events/arch/x86/alderlake/cache.json b/tools/pe= rf/pmu-events/arch/x86/alderlake/cache.json index a20e19738046..04c53035e967 100644 --- a/tools/perf/pmu-events/arch/x86/alderlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/alderlake/cache.json @@ -1050,6 +1050,28 @@ "UMask": "0x3", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.COREWB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10008", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -1094,6 +1116,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -1160,6 +1204,28 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -1215,6 +1281,17 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that have any type of respons= e.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.SWPF_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x14000", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by the L3 = cache.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/alderlake/memory.json b/tools/p= erf/pmu-events/arch/x86/alderlake/memory.json index fa15f5797bed..c06507a40bde 100644 --- a/tools/perf/pmu-events/arch/x86/alderlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/alderlake/memory.json @@ -253,6 +253,17 @@ "UMask": "0x2", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -264,6 +275,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -297,6 +330,17 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -330,6 +374,17 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.SWPF_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784004000", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were not supplied by the= L3 cache.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/alderlake/other.json b/tools/pe= rf/pmu-events/arch/x86/alderlake/other.json index a8b23e92408c..ae3a6630bd72 100644 --- a/tools/perf/pmu-events/arch/x86/alderlake/other.json +++ b/tools/perf/pmu-events/arch/x86/alderlake/other.json @@ -55,116 +55,6 @@ "UMask": "0x1", "Unit": "cpu_atom" }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.COREWB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10008", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3,4,5", @@ -209,92 +99,6 @@ "UMask": "0x1", "Unit": "cpu_core" }, - { - "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that have any type of respons= e.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.SWPF_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x14000", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.SWPF_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784004000", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "Deprecated": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.COUNT", - "Invert": "1", - "SampleAfterValue": "100003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", - "Counter": "0,1,2,3,4,5,6,7", - "Deprecated": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.CYCLES", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state. For Tremont, UMWAIT and TPAUSE will onl= y put the CPU into C0.1 activity state (not C0.2 activity state)", - "Counter": "0,1,2,3,4,5", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/alderlake/pipeline.json b/tools= /perf/pmu-events/arch/x86/alderlake/pipeline.json index f5bf0816f190..f08c1d6a99ba 100644 --- a/tools/perf/pmu-events/arch/x86/alderlake/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/alderlake/pipeline.json @@ -1213,8 +1213,9 @@ "Unit": "cpu_atom" }, { - "BriefDescription": "Counts the number of machine clears that flus= h the pipeline and restart the machine with the use of microcode due to SMC= , MEMORY_ORDERING, FP_ASSISTS, PAGE_FAULT, DISAMBIGUATION, and FPC_VIRTUAL_= TRAP.", + "BriefDescription": "This event is deprecated.", "Counter": "0,1,2,3,4,5", + "Deprecated": "1", "EventCode": "0xc3", "EventName": "MACHINE_CLEARS.SLOW", "SampleAfterValue": "20003", @@ -1289,6 +1290,70 @@ "UMask": "0x2", "Unit": "cpu_core" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "Deprecated": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.COUNT", + "Invert": "1", + "SampleAfterValue": "100003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", + "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.CYCLES", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state. For Tremont, UMWAIT and TPAUSE will onl= y put the CPU into C0.1 activity state (not C0.2 activity state)", + "Counter": "0,1,2,3,4,5", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts the number of issue slots not consumed= by the backend due to a micro-sequencer (MS) scoreboard, which stalls the = front-end from issuing from the UROM until a specified older uop retires.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 56d5fc419acf..881f418137fd 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -1,5 +1,5 @@ Family-model,Version,Filename,EventType -GenuineIntel-6-(97|9A|B7|BA|BF),v1.28,alderlake,core +GenuineIntel-6-(97|9A|B7|BA|BF),v1.29,alderlake,core GenuineIntel-6-BE,v1.28,alderlaken,core GenuineIntel-6-C[56],v1.07,arrowlake,core GenuineIntel-6-(1C|26|27|35|36),v5,bonnell,core --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id C20952E3373 for ; Wed, 26 Mar 2025 00:05:19 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947522; cv=none; b=Yge+2/xNsoBrfOlnQ1V/Q/XODsgZ46z6UvCFJwud1ljK8Z1TBQ4aNJZGb0JWzAXLg6jV9HNTbtllAGDeLpQcM+eU0o8oiTg6KP2Q6PrzmrRWd5VQ7KoA5CWj5hockQPGfLiLSHG6xTWOYQ81SP1WD0y6FhfJ81tjPei9d1SeVz4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947522; c=relaxed/simple; bh=s0JRgLoxxuvunaHY7JyhOI/QMjmxBJL1LP9dIRrQ/aA=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=hTTHT95QAv4tiUZfnLzAPlm7dUXwafocYDTuMqLMI7s9TxGApSLzGkqam2IRQFwdYXOv1SD27Wx+VPiAsXu7ByYTtpXQ2LIxcv09xe6xrSoD/KdKpYaPh3HOgQM2XYvgH8AzMbqNoGXlWh1zK+SLJwAG/YCWfgOrSeSZS5vVkSk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=HJGcqSZN; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="HJGcqSZN" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff798e8c93so10555527a91.2 for ; Tue, 25 Mar 2025 17:05:19 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947519; x=1743552319; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=aGWGBvU/x5QSdgsAS6s6D7Oyp5KpZml9VkSt87fTBIw=; b=HJGcqSZNhTXdlsq7hRb+/LSiknbXZIfupjkOvzD2T7cHUfj6KgpJ+QTnuUR5xKyB3I SOmFee+67c01DujIqgl79B9Q/yTfBR8eVavRzuHRVo6Ok88nF+/fkLoFUF+NPQEVQhzI TtHWsFml/lXReYeSyQf5h8TgUvpgM0u0RN2VbyJo0O0cFoN2WtHEZg+sDujSBWAR6Jm4 sDu7OTJ6PMRPjzkgGFPAC7eA1xFwNDNDdj9WHAULxtC8XwxCPV8fPC84ARIQLG5JXkha vicwFZUQMtahK5kZHrDnAzMMJFHF1mo/1iLP9zmYOtRtQPVkwd8pUsK/PGx4mEiTJ5mj ugjA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947519; x=1743552319; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=aGWGBvU/x5QSdgsAS6s6D7Oyp5KpZml9VkSt87fTBIw=; b=TenfVku6cex3yDJ4TOKWOYo04cWgTWBDEER4TJbLmhICJCwUSGWsUpr4dVO+dv6gEO sEnAWwD56QVemUS2v8u9pYPPv9rJFAR0BD/fmZq0jgQS5P0Hnk/Edfn9Mblh6XIdTC9n doZp3d4jgQWljV1/ql7pTIUC0JjBXg9bwLfjP2ElukYxcD65Os5JjX57JmmoLtp931xo 2N4h0Md3fFkrZSXDVyZl/wjizM6gQRT251FG5fK9Nw1ojr2eeYVuN7rAoFbjxyoUXDSg b0RtHdTh/q/WZ1mgRcrOuD7t/NpwhgtJU3uWa6vNS0Vh2mg6yXEVCQM23Gj+M8rcugms IeiQ== X-Forwarded-Encrypted: i=1; AJvYcCVTE7/hxPR/8wOJdYeF42UY04rUDOBCJsqCeiKoqW2hgpzJgreBxAtiYrFOO733XQcaXtMHV2oGdBWEsck=@vger.kernel.org X-Gm-Message-State: AOJu0Yy3aN9MHv3tvCixTPK1jszrvO3eN001+Xc0Hq5lWNbB2OqdRVLK iGJqrP0K1T3Uug8WtnMdsHZW1u1LrvkoLhrAc43uFweLPHCKOHCzqbpcAwWggjaPEYbZYv2H75J 3MiUCgA== X-Google-Smtp-Source: AGHT+IH8r+oHk9yRWt/mYha62PxF4bSI8qeNl2jt9z/IDq5ZObjpVNzM7fb2MDiT1Z+6fmh5qurGS7ePI8ZX X-Received: from pgau25.prod.google.com ([2002:a05:6a02:2d99:b0:af2:211e:938e]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a20:7488:b0:1f3:194b:30b0 with SMTP id adf61e73a8af0-1fe42f25e29mr30484136637.5.1742947518974; Tue, 25 Mar 2025 17:05:18 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:00 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-3-irogers@google.com> Subject: [PATCH v3 02/35] perf vendor events: Update AlderlakeN events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.28 to v1.29. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/alderlaken/adln-metrics.json | 54 +++++----- .../pmu-events/arch/x86/alderlaken/cache.json | 50 ++++++++++ .../arch/x86/alderlaken/memory.json | 40 ++++++++ .../pmu-events/arch/x86/alderlaken/other.json | 98 ------------------- .../arch/x86/alderlaken/pipeline.json | 11 ++- tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- 6 files changed, 128 insertions(+), 127 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/alderlaken/adln-metrics.json b/= tools/perf/pmu-events/arch/x86/alderlaken/adln-metrics.json index ad04b1e3881e..ce93648043ef 100644 --- a/tools/perf/pmu-events/arch/x86/alderlaken/adln-metrics.json +++ b/tools/perf/pmu-events/arch/x86/alderlaken/adln-metrics.json @@ -75,7 +75,7 @@ "MetricExpr": "tma_core_bound", "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", - "MetricThreshold": "(tma_allocation_restriction >0.10) & ((tma_cor= e_bound >0.10) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { @@ -84,7 +84,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.ALL / (5 * CPU_CLK_UNHALTED.CORE)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound >0.10)", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%" @@ -95,7 +95,7 @@ "MetricExpr": "(5 * CPU_CLK_UNHALTED.CORE - (TOPDOWN_FE_BOUND.ALL = + TOPDOWN_BE_BOUND.ALL + TOPDOWN_RETIRING.ALL)) / (5 * CPU_CLK_UNHALTED.COR= E)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation >0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%" @@ -105,7 +105,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_DETECT / (5 * CPU_CLK_UNHAL= TED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "MetricThreshold": "(tma_branch_detect >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%" }, @@ -114,7 +114,7 @@ "MetricExpr": "TOPDOWN_BAD_SPECULATION.MISPREDICT / (5 * CPU_CLK_U= NHALTED.CORE)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts >0.05) & ((tma_bad_spe= culation >0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -123,7 +123,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_RESTEER / (5 * CPU_CLK_UNHA= LTED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", - "MetricThreshold": "(tma_branch_resteer >0.05) & ((tma_ifetch_late= ncy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -131,7 +131,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.CISC / (5 * CPU_CLK_UNHALTED.CORE)= ", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc >0.05) & ((tma_ifetch_bandwidth >0.1= 0) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -139,7 +139,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS / (5 * CPU_CLK_= UNHALTED.CORE)", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound >0.10) & ((tma_backend_bound >= 0.10))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -148,7 +148,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.DECODE / (5 * CPU_CLK_UNHALTED.COR= E)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", - "MetricThreshold": "(tma_decode >0.05) & ((tma_ifetch_bandwidth >0= .10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -156,7 +156,7 @@ "MetricExpr": "TOPDOWN_BAD_SPECULATION.FASTNUKE / (5 * CPU_CLK_UNH= ALTED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", - "MetricThreshold": "(tma_fast_nuke >0.05) & ((tma_machine_clears >= 0.05) & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -165,7 +165,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.ALL / (5 * CPU_CLK_UNHALTED.CORE)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound >0.20)", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%" }, @@ -174,7 +174,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.ICACHE / (5 * CPU_CLK_UNHALTED.COR= E)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -182,7 +182,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH / (5 * CPU_CLK_= UNHALTED.CORE)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", - "MetricThreshold": "(tma_ifetch_bandwidth >0.10) & ((tma_frontend_= bound >0.20))", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -191,7 +191,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_LATENCY / (5 * CPU_CLK_UN= HALTED.CORE)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", - "MetricThreshold": "(tma_ifetch_latency >0.15) & ((tma_frontend_bo= und >0.20))", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -473,7 +473,7 @@ "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "CPU_CLK_UNHALTED.CORE_P / CPU_CLK_UNHALTED.CORE", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))" + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", @@ -506,7 +506,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.ITLB / (5 * CPU_CLK_UNHALTED.CORE)= ", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses >0.05) & ((tma_ifetch_latency= >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -514,7 +514,7 @@ "MetricExpr": "TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS / (5 * CPU_C= LK_UNHALTED.CORE)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears >0.05) & ((tma_bad_specula= tion >0.15))", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -523,7 +523,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.MEM_SCHEDULER / (5 * CPU_CLK_UNHAL= TED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", - "MetricThreshold": "(tma_mem_scheduler >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { @@ -531,7 +531,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER / (5 * CPU_CLK_U= NHALTED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", - "MetricThreshold": "(tma_non_mem_scheduler >0.10) & ((tma_resource= _bound >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { @@ -539,7 +539,7 @@ "MetricExpr": "TOPDOWN_BAD_SPECULATION.NUKE / (5 * CPU_CLK_UNHALTE= D.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", - "MetricThreshold": "(tma_nuke >0.05) & ((tma_machine_clears >0.05)= & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -547,7 +547,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.OTHER / (5 * CPU_CLK_UNHALTED.CORE= )", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", - "MetricThreshold": "(tma_other_fb >0.05) & ((tma_ifetch_bandwidth = >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -555,7 +555,7 @@ "MetricExpr": "TOPDOWN_FE_BOUND.PREDECODE / (5 * CPU_CLK_UNHALTED.= CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", - "MetricThreshold": "(tma_predecode >0.05) & ((tma_ifetch_bandwidth= >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -563,7 +563,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.REGISTER / (5 * CPU_CLK_UNHALTED.C= ORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", - "MetricThreshold": "(tma_register >0.10) & ((tma_resource_bound >0= .20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { @@ -571,7 +571,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.REORDER_BUFFER / (5 * CPU_CLK_UNHA= LTED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", - "MetricThreshold": "(tma_reorder_buffer >0.10) & ((tma_resource_bo= und >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { @@ -579,7 +579,7 @@ "MetricExpr": "tma_backend_bound - tma_core_bound", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", - "MetricThreshold": "(tma_resource_bound >0.20) & ((tma_backend_bou= nd >0.10))", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -589,7 +589,7 @@ "MetricExpr": "TOPDOWN_RETIRING.ALL / (5 * CPU_CLK_UNHALTED.CORE)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "(tma_retiring >0.75)", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%" }, @@ -598,7 +598,7 @@ "MetricExpr": "TOPDOWN_BE_BOUND.SERIALIZATION / (5 * CPU_CLK_UNHAL= TED.CORE)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", - "MetricThreshold": "(tma_serialization >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" } ] diff --git a/tools/perf/pmu-events/arch/x86/alderlaken/cache.json b/tools/p= erf/pmu-events/arch/x86/alderlaken/cache.json index fd9ed58c2f90..605d56311dfc 100644 --- a/tools/perf/pmu-events/arch/x86/alderlaken/cache.json +++ b/tools/perf/pmu-events/arch/x86/alderlaken/cache.json @@ -396,6 +396,26 @@ "SampleAfterValue": "1000003", "UMask": "0x6" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.COREWB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10008", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -436,6 +456,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -476,6 +506,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -516,6 +556,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that have any type of respons= e.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.SWPF_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x14000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by the L3 = cache.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/alderlaken/memory.json b/tools/= perf/pmu-events/arch/x86/alderlaken/memory.json index 3b46b048dfb2..06eca0a45c18 100644 --- a/tools/perf/pmu-events/arch/x86/alderlaken/memory.json +++ b/tools/perf/pmu-events/arch/x86/alderlaken/memory.json @@ -56,6 +56,16 @@ "SampleAfterValue": "20003", "UMask": "0x2" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -66,6 +76,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -86,6 +106,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5", @@ -106,6 +136,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5", + "EventCode": "0xB7", + "EventName": "OCR.SWPF_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x784004000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were not supplied by the= L3 cache.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/alderlaken/other.json b/tools/p= erf/pmu-events/arch/x86/alderlaken/other.json index f8c21b7f8f40..0ebcb3e20e1d 100644 --- a/tools/perf/pmu-events/arch/x86/alderlaken/other.json +++ b/tools/perf/pmu-events/arch/x86/alderlaken/other.json @@ -8,76 +8,6 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.COREWB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10008", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3,4,5", @@ -107,33 +37,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that have any type of respons= e.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.SWPF_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x14000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache software prefetches whic= h include T0/T1/T2 and NTA (except PREFETCHW) that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5", - "EventCode": "0xB7", - "EventName": "OCR.SWPF_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x784004000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state. For Tremont, UMWAIT and TPAUSE will onl= y put the CPU into C0.1 activity state (not C0.2 activity state)", - "Counter": "0,1,2,3,4,5", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4" } ] diff --git a/tools/perf/pmu-events/arch/x86/alderlaken/pipeline.json b/tool= s/perf/pmu-events/arch/x86/alderlaken/pipeline.json index 713ebc21cec0..0cc8cd203fc3 100644 --- a/tools/perf/pmu-events/arch/x86/alderlaken/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/alderlaken/pipeline.json @@ -399,8 +399,9 @@ "UMask": "0x20" }, { - "BriefDescription": "Counts the number of machine clears that flus= h the pipeline and restart the machine with the use of microcode due to SMC= , MEMORY_ORDERING, FP_ASSISTS, PAGE_FAULT, DISAMBIGUATION, and FPC_VIRTUAL_= TRAP.", + "BriefDescription": "This event is deprecated.", "Counter": "0,1,2,3,4,5", + "Deprecated": "1", "EventCode": "0xc3", "EventName": "MACHINE_CLEARS.SLOW", "SampleAfterValue": "20003", @@ -423,6 +424,14 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state. For Tremont, UMWAIT and TPAUSE will onl= y put the CPU into C0.1 activity state (not C0.2 activity state)", + "Counter": "0,1,2,3,4,5", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4" + }, { "BriefDescription": "Counts the number of issue slots not consumed= by the backend due to a micro-sequencer (MS) scoreboard, which stalls the = front-end from issuing from the UROM until a specified older uop retires.", "Counter": "0,1,2,3,4,5", diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 881f418137fd..0ef31b65f8df 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -1,6 +1,6 @@ Family-model,Version,Filename,EventType GenuineIntel-6-(97|9A|B7|BA|BF),v1.29,alderlake,core -GenuineIntel-6-BE,v1.28,alderlaken,core +GenuineIntel-6-BE,v1.29,alderlaken,core GenuineIntel-6-C[56],v1.07,arrowlake,core GenuineIntel-6-(1C|26|27|35|36),v5,bonnell,core GenuineIntel-6-(3D|47),v30,broadwell,core --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f201.google.com (mail-pl1-f201.google.com [209.85.214.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 499434A32 for ; Wed, 26 Mar 2025 00:05:27 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.201 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947535; cv=none; b=sAYjXjIp94qwq7yDBBMl6hvyDO5/2FERwN+GysSUyb1qWj8ga6wK6dWwKK0sctpWuNWsoTte+eQ/b9DS6Xij0KjedI77Cfa5u62cG4YjHLw0jlJVbUUzH3YU41I3zmdMDXYm0yadQEyPzh4TsblVhr7YBSt9uBqN5ACxH0lQJH4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947535; c=relaxed/simple; bh=JTmMzTLA5iwql0ORnNF497cG4jDt1OgdTobDIjGpXXE=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=NTpVjtode+DBDLF2r1KKangYFWV4T7rULY/AVrlAPuYyp1itaW+HcHKQfHCI2b1lbFOzp+Ccx6W2y50/2Hhcfc57LnuX2CR3qn4yZQU1A7K42dPl4myvi1mMILMQlMYNCpJNmuIbPp0Xi323nS+yPMIF7q1aXnhI5Y7W34FsoFc= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=zTZOylyr; arc=none smtp.client-ip=209.85.214.201 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="zTZOylyr" Received: by mail-pl1-f201.google.com with SMTP id d9443c01a7336-227ed471999so7984215ad.3 for ; Tue, 25 Mar 2025 17:05:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947527; x=1743552327; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=dhL/uxaj6AOH2HIoQYYxqxbUvtYQ774Ps+abRZ0TmTM=; b=zTZOylyrFvqij7W6JGb0cRI20WG2LSLmEtiyPaAFsEhlg43SnhDxc93VDdau5NgQY+ oGTL45at5ZD+iXNQWn0jKyqny6Gbwza2k4QYNRJiSh8+8xOBFkDFcF7pPft2/ZHyWUyE q3xJHlGLHTb3nsMBeoyAAQQJZ8G1sTK7o5iTm68fEnGg1dQ5JWdHo2nypUctGP72b6Mf OfU0EGCmN8ASLUFCVt1wo34wrVF1ihYuHRvmKf8IMfLk8vyHZXjlUZhqxdxirDQJ/u6U 2US/gCE/yi6oHJO5x5W20kCFQzETv9Hk/nQ/nrQZtIryLevoX/YCHwAObM5DDwY4jm+z Svxg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947527; x=1743552327; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=dhL/uxaj6AOH2HIoQYYxqxbUvtYQ774Ps+abRZ0TmTM=; b=kkj8oXOk5PWVunyOsa3HTuXNDuvsMXBRTjfdG2jdu9IG8BT9CbEuJ54fIsl/g0bDgK iUiUEd/8LU+alfSWT8DfOyKN4g8qDVT+nzQDzPjgtPTsJTPnWaKj9scCZER1P+YqTYiB o0VQpJnZIkKz7uoUNvX2HtJGlrAxPGPHFv0lD2uxgHYJiPPuA4m187ClG7n3C0ovOou+ gQZAgKCB41GZBHCaHoD6WaTrUAjw/bRqxxFW9yaUOgbOYF5/2VmzXCLJbrtWil8i6RT3 LJHf3uoX9JmQuQPtdiJqRBfazPlYI++7L+FsV1MMZtnpeIzOWHeOKwgdWcgnrg7G8PyJ OYBw== X-Forwarded-Encrypted: i=1; AJvYcCVINK/1qx3eSBsdkHKvps6kq35ZLDGRb/DV3Aa1s1m2i3UARgV002yLEZuSHiNpS5QQz9oYWW2I0VAiZH0=@vger.kernel.org X-Gm-Message-State: AOJu0YyVGz4D/g50uc1ekK2cUPg7WTL2EBkxRsCI2k8QSI52H44NK6Mr Cdv+sa2v7E3SUPaGzME8jQ/mP5Jz14aJBUkRu6VtNQuc5ANkUES/iDowr/QSCZWK0RNuSvXDDLZ 7MVJw0w== X-Google-Smtp-Source: AGHT+IHTs1j4Dny1Dj6I8w0dP2WVdzSw2CdMUw7XNoFWsMm8ebNnMObHUTZJSDbAwjwN9jgpuhzdRaZQEqo2 X-Received: from pfbna8.prod.google.com ([2002:a05:6a00:3e08:b0:734:cc8:a107]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:902:d4ca:b0:223:4537:65b1 with SMTP id d9443c01a7336-22780e10a22mr283356225ad.36.1742947526741; Tue, 25 Mar 2025 17:05:26 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:01 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-4-irogers@google.com> Subject: [PATCH v3 03/35] perf vendor events: Update arrowlake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.07 to v1.08. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/arrowlake/arl-metrics.json | 562 +++++++++--------- .../pmu-events/arch/x86/arrowlake/cache.json | 148 +++++ .../pmu-events/arch/x86/arrowlake/memory.json | 11 + .../pmu-events/arch/x86/arrowlake/other.json | 193 ------ .../arch/x86/arrowlake/pipeline.json | 163 ++++- tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- 6 files changed, 606 insertions(+), 473 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/arrowlake/arl-metrics.json b/to= ols/perf/pmu-events/arch/x86/arrowlake/arl-metrics.json index 7ddb89dd1871..2d26d57aff34 100644 --- a/tools/perf/pmu-events/arch/x86/arrowlake/arl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/arrowlake/arl-metrics.json @@ -75,7 +75,7 @@ "MetricExpr": "tma_core_bound", "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", - "MetricThreshold": "(tma_allocation_restriction >0.10) & ((tma_cor= e_bound >0.10) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -85,7 +85,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALL_P@ / (8 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound >0.10)", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%", @@ -97,7 +97,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.ALL_P@ / (8 * cpu_= atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation >0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%", @@ -108,7 +108,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_DETECT@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "MetricThreshold": "(tma_branch_detect >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -118,7 +118,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MISPREDICT@ / (8 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts >0.05) & ((tma_bad_spe= culation >0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -128,7 +128,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_RESTEER@ / (8 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", - "MetricThreshold": "(tma_branch_resteer >0.05) & ((tma_ifetch_late= ncy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -137,7 +137,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.CISC@ / (8 * cpu_atom@CPU= _CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc >0.05) & ((tma_ifetch_bandwidth >0.1= 0) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -146,7 +146,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS@ / (8 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound >0.10) & ((tma_backend_bound >= 0.10))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -156,7 +156,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.DECODE@ / (8 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", - "MetricThreshold": "(tma_decode >0.05) & ((tma_ifetch_bandwidth >0= .10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -165,7 +165,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.FASTNUKE@ / (8 * c= pu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", - "MetricThreshold": "(tma_fast_nuke >0.05) & ((tma_machine_clears >= 0.05) & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -175,7 +175,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ALL@ / (8 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound >0.20)", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -185,7 +185,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ICACHE@ / (8 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -194,7 +194,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH@ / (8 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", - "MetricThreshold": "(tma_ifetch_bandwidth >0.10) & ((tma_frontend_= bound >0.20))", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -204,7 +204,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_LATENCY@ / (8 * = cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", - "MetricThreshold": "(tma_ifetch_latency >0.15) & ((tma_frontend_bo= und >0.20))", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -590,7 +590,7 @@ "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "cpu_atom@CPU_CLK_UNHALTED.CORE_P@ / cpu_atom@CPU_CL= K_UNHALTED.CORE@", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9", "Unit": "cpu_atom" }, { @@ -629,7 +629,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ITLB_MISS@ / (8 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses >0.05) & ((tma_ifetch_latency= >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -638,7 +638,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS@ / = (8 * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears >0.05) & ((tma_bad_specula= tion >0.15))", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -648,7 +648,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.MEM_SCHEDULER@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", - "MetricThreshold": "(tma_mem_scheduler >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -657,7 +657,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER@ / (8 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", - "MetricThreshold": "(tma_non_mem_scheduler >0.10) & ((tma_resource= _bound >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -666,7 +666,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.NUKE@ / (8 * cpu_a= tom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", - "MetricThreshold": "(tma_nuke >0.05) & ((tma_machine_clears >0.05)= & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -675,7 +675,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.OTHER@ / (8 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", - "MetricThreshold": "(tma_other_fb >0.05) & ((tma_ifetch_bandwidth = >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -684,7 +684,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.PREDECODE@ / (8 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", - "MetricThreshold": "(tma_predecode >0.05) & ((tma_ifetch_bandwidth= >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -693,7 +693,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REGISTER@ / (8 * cpu_atom= @CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", - "MetricThreshold": "(tma_register >0.10) & ((tma_resource_bound >0= .20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -702,7 +702,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REORDER_BUFFER@ / (8 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", - "MetricThreshold": "(tma_reorder_buffer >0.10) & ((tma_resource_bo= und >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -711,7 +711,7 @@ "MetricExpr": "tma_backend_bound - tma_core_bound", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", - "MetricThreshold": "(tma_resource_bound >0.20) & ((tma_backend_bou= nd >0.10))", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -722,7 +722,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_RETIRING.ALL@ / (8 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "(tma_retiring >0.75)", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -732,7 +732,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.SERIALIZATION@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", - "MetricThreshold": "(tma_serialization >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -744,7 +744,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "cpu_core@UOPS_DISPATCHED.ALU@ / (6 * tma_info_threa= d_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -757,13 +757,13 @@ "MetricExpr": "78 * cpu_core@ASSISTS.ANY@ / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_threa= d_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -774,7 +774,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-be\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -786,18 +786,18 @@ { "BriefDescription": "This category represents fraction of slots wa= sted due to incorrect speculations", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-bad\\-spec / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-bad\\-spec@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20", @@ -814,11 +814,11 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_= split_loads + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_capacity + tma_l1_latency_dependency + tma_lock_latency + tma_split_= loads + tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", - "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_mem_b= andwidth, tma_sq_full", + "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_= system_dram_bw_use, tma_mem_bandwidth, tma_sq_full", "Unit": "cpu_core" }, { @@ -826,22 +826,22 @@ "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity= + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_sto= re_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_latency_c= apacity / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma= _l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtl= b_load + tma_fb_full + tma_l1_latency_capacity + tma_l1_latency_dependency = + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bou= nd * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_fb_ful= l + tma_l1_latency_capacity + tma_l1_latency_dependency + tma_lock_latency = + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bou= nd / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_sto= re_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_= split_stores + tma_store_latency + tma_streaming_stores)) + tma_memory_boun= d * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_f= alse_sharing + tma_split_stores + tma_store_latency + tma_streaming_stores)= ))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", - "MetricThreshold": "(tma_bottleneck_cache_memory_latency > 20)", + "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Latency related bottlenecks. Related metrics: tma_l3_hit_latency, tma_= mem_latency", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy", + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: ", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_= clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_br= anch_mispredicts) / (tma_mispredicts_resteers + tma_clears_resteers + tma_u= nknown_branches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_reste= ers + tma_ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth *= tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_co= de", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cl= ears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_bran= ch_mispredicts) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unk= nown_branches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_miss= es + tma_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * t= ma_ms / (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code= ", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", @@ -849,7 +849,7 @@ }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_s= witches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_rest= eers * tma_other_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_r= esteers + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses= + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_= dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ls= d + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other= _nukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cp= u_core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / = (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_mic= rocode_sequencer / (tma_microcode_sequencer + tma_few_uops_instructions) * = (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_restee= rs * tma_other_mispredicts / tma_branch_mispredicts) / (tma_clears_resteers= + tma_mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers= + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_m= s_switches) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite = + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_bra= nch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_n= ukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cpu_= core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / (t= ma_divider + tma_ports_utilization + tma_serializing_operation) + tma_micro= code_sequencer / (tma_microcode_sequencer + tma_few_uops_instructions) * (t= ma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -861,7 +861,7 @@ "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fw= d_blk)) + tma_memory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bo= und + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_dtlb_store / (= tma_dtlb_store + tma_false_sharing + tma_split_stores + tma_store_latency += tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", - "MetricThreshold": "(tma_bottleneck_memory_data_tlbs > 20)", + "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", "PublicDescription": "Total pipeline cost of Memory Address Transl= ation related bottlenecks (data-side TLBs). Related metrics: tma_dtlb_load,= tma_dtlb_store", "Unit": "cpu_core" }, @@ -870,13 +870,13 @@ "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", - "MetricThreshold": "(tma_bottleneck_memory_synchronization > 10)", + "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -888,12 +888,12 @@ "MetricExpr": "100 - (tma_bottleneck_big_code + tma_bottleneck_ins= truction_fetch_bw + tma_bottleneck_mispredictions + tma_bottleneck_cache_me= mory_bandwidth + tma_bottleneck_cache_memory_latency + tma_bottleneck_memor= y_data_tlbs + tma_bottleneck_memory_synchronization + tma_bottleneck_comput= e_bound_est + tma_bottleneck_irregular_overhead + tma_bottleneck_branching_= overhead + tma_bottleneck_useful_work)", "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", - "MetricThreshold": "(tma_bottleneck_other_bottlenecks > 20)", + "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls.", "Unit": "cpu_core" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (cpu_core@BR_INST_RETIRED.ALL= _BRANCHES@ + 2 * cpu_core@BR_INST_RETIRED.NEAR_CALL@ + cpu_core@INST_RETIRE= D.NOP@) / tma_info_thread_slots - tma_microcode_sequencer / (tma_microcode_= sequencer + tma_few_uops_instructions) * (tma_assists / tma_microcode_seque= ncer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -902,7 +902,7 @@ }, { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "cpu_core@topdown\\-br\\-mispredict@ / (cpu_core@top= down\\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-re= tiring@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -916,26 +916,26 @@ "MetricExpr": "cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_= thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C01@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C02@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -944,8 +944,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -954,99 +954,99 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@frontend_retired.l1i_miss@R / tma_info_thread_clks - tma_code_l2_miss)", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@FRONTEND_RETIRED.L1I_MISS@R / tma_info_thread_clks - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@front= end_retired.l2_miss@R / tma_info_thread_clks", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@FRONT= END_RETIRED.L2_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the (first level) ITLB was missed by instructions fetches, th= at later on hit in second-level TLB (STLB)", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@frontend_retired.itlb_miss@R / tma_info_thread_clks - tma_code_stlb_mis= s)", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@FRONTEND_RETIRED.ITLB_MISS@R / tma_info_thread_clks - tma_code_stlb_mis= s)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates the fraction of cycles = where the Second-level TLB (STLB) was missed by instruction fetches, perfor= ming a hardware page walk", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@fro= ntend_retired.stlb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@FRO= NTEND_RETIRED.STLB_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@ / (cpu_core@ITLB_MISSES.= WALK_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_4K@ / (cpu_core@ITLB_MISSES.WAL= K_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@br_misp_retired.cond_ntaken_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@BR_MISP_RETIRED.COND_NTAKEN_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_nt_mispredicts", - "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by backward-taken conditional branche= s", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@ * cpu= _core@br_misp_retired.cond_taken_bwd_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by backward-taken conditional branche= s.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@ * cpu= _core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_bwd_mispredicts", - "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & tma_branc= h_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & (tma_bran= ch_mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by forward-taken conditional branches= ", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@ * cpu= _core@br_misp_retired.cond_taken_fwd_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by forward-taken conditional branches= .", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@ * cpu= _core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_fwd_mispredicts", - "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & tma_branc= h_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & (tma_bran= ch_mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ *= cpu_core@mem_load_l3_hit_retired.xsnp_miss@R, cpu_core@MEM_LOAD_L3_HIT_RET= IRED.XSNP_MISS@ * (27 * tma_info_system_core_frequency) - 3 * tma_info_syst= em_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_miss@R else= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * (27 * tma_info_system_core_f= requency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM@ * cpu_core@mem_load_l3_hit_retired.xsnp_hitm@R, cpu= _core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * (28 * tma_info_system_core_frequ= ency) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit= _retired.xsnp_hitm@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * (28= * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequency)) *= (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS= @ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * min(= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R, 24 * tma_info_system_core_fre= quency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * min(cpu_core@MEM_LO= AD_L3_HIT_RETIRED.XSNP_HITM@R, 25 * tma_info_system_core_frequency)) * (1 += cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2= ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1057,17 +1057,17 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@= * cpu_core@mem_load_l3_hit_retired.xsnp_no_fwd@R, cpu_core@MEM_LOAD_L3_HIT= _RETIRED.XSNP_NO_FWD@ * (27 * tma_info_system_core_frequency) - 3 * tma_inf= o_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_no_fw= d@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * (27 * tma_info_sys= tem_core_frequency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@M= EM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * cpu_core@mem_load_l3_hit_retired.xsnp_fw= d@R, cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (28 * tma_info_system_cor= e_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load= _l3_hit_retired.xsnp_fwd@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ = * (28 * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequenc= y)) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1= _MISS@ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * mi= n(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R, 24 * tma_info_system_core= _frequency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * min(cpu_core@MEM= _LOAD_L3_HIT_RETIRED.XSNP_FWD@R, 25 * tma_info_system_core_frequency)) * (1= + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ /= 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_contested_accesses, tma= _machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1076,7 +1076,7 @@ "MetricExpr": "cpu_core@ARITH.DIV_ACTIVE@ / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1086,7 +1086,7 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.MEM@ / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1097,7 +1097,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1106,28 +1106,28 @@ "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / tma_in= fo_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * cp= u_core@mem_inst_retired.stlb_hit_loads@R, cpu_core@MEM_INST_RETIRED.STLB_HI= T_LOADS@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_loads@R else cpu_co= re@MEM_INST_RETIRED.STLB_HIT_LOADS@ * 7) / tma_info_thread_clks + tma_load_= stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * min(cpu= _core@MEM_INST_RETIRED.STLB_HIT_LOADS@R, 7) / tma_info_thread_clks + tma_lo= ad_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * c= pu_core@mem_inst_retired.stlb_hit_stores@R, cpu_core@MEM_INST_RETIRED.STLB_= HIT_STORES@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_stores@R else cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * 7) / tma_info_thread_clks + tma_= store_stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * min(cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@R, 7) / tma_info_thread_clks + tma_= store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1136,7 +1136,7 @@ "MetricExpr": "28 * tma_info_system_core_frequency * cpu_core@OCR.= DEMAND_RFO.L3_HIT.SNOOP_HITM@ / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "(tma_false_sharing > 0.05) & ((tma_store_bound= > 0.2) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1147,7 +1147,7 @@ "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1158,18 +1158,18 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Rel= ated metrics: tma_dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_= botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_inst_mix_ipt= b, tma_lcp. Sample with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEN= D_RETIRED.LATENCY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_= dsb_switches, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_miss= es, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = CPU was stalled due to Frontend latency issues", - "MetricExpr": "topdown\\-fetch\\-lat / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-fetch\\-lat@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend= _bound_group", "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1179,7 +1179,7 @@ "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;= tma_issueD0", "MetricName": "tma_few_uops_instructions", "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_o= perations > 0.1", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring instructions that that are decoder into two or more= uops. This highly-correlates with the number of uops in such instructions", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring instructions that that are decoder into two or more= uops. This highly-correlates with the number of uops in such instructions.= Related metrics: tma_decoder0_alone", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1189,7 +1189,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1199,16 +1199,16 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "cpu_core@ARITH.FPDIV_ACTIVE@ / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1217,8 +1217,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.SCALAR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utili= zed_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1227,8 +1227,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, = tma_int_vector_256b, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1237,8 +1237,8 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b,= tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1247,15 +1247,15 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR\\,umask\\=3D0= x30@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b,= tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots wh= ere the processor's Frontend undersupplies its Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-fe\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-fe\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;Default;PGO;TmaL1;TopdownL1;tma_L1_group= ", "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", @@ -1265,23 +1265,23 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.MACRO_= FUSED@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "cpu_core@topdown\\-heavy\\-ops@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1290,26 +1290,26 @@ "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / tma_info_thread_clks= ", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= ", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@br_misp_retired.indirect_call_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= .", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_call_mispredicts", - "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions", - "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@br_misp_retired.indirect_cost@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@br_misp_retired.indirect_call_cost@R) / tma_info_thread= _clks, 0)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions.= ", + "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_COST@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R) / tma_info_thread= _clks, 0)", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_jump_mispredicts", - "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1322,7 +1322,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_NTAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", @@ -1330,29 +1330,29 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional backward-taken branches (lower number means higher occurrence rate)= ", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional backward-taken branches (lower number means higher occurrence rate)= .", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN_BWD@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_bwd", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional forward-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional forward-taken branches (lower number means higher occurrence rate).= ", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN_FWD@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_fwd", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.INDIRECT@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.RET@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -1376,7 +1376,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -1385,7 +1385,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -1394,10 +1394,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: ", "Unit": "cpu_core" }, { @@ -1463,12 +1464,12 @@ "MetricExpr": "(cpu_core@FP_ARITH_DISPATCHED.V0@ + cpu_core@FP_ARI= TH_DISPATCHED.V1@ + cpu_core@FP_ARITH_DISPATCHED.V2@ + cpu_core@FP_ARITH_DI= SPATCHED.V3@) / (4 * tma_info_thread_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)", + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n).", "Unit": "cpu_core" }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp", "Unit": "cpu_core" @@ -1483,15 +1484,15 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired DSB misses", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= frontend_retired.any_dsb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= FRONTEND_RETIRED.ANY_DSB_MISS@R / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat", "MetricName": "tma_info_frontend_dsb_switches_ret", "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05", @@ -1499,7 +1500,7 @@ }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D1@", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc", "Unit": "cpu_core" @@ -1549,7 +1550,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired operations that invoke th= e Microcode Sequencer", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@fron= tend_retired.ms_flows@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@FRON= TEND_RETIRED.MS_FLOWS@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq", "MetricName": "tma_info_frontend_ms_latency_ret", "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05", @@ -1564,21 +1565,21 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node", + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node.", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired branches who got branch a= ddress clears", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@frontend_retired.unknown_branch@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@FRONTEND_RETIRED.UNKNOWN_BRANCH@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat", "MetricName": "tma_info_frontend_unknown_branches_ret", "Unit": "cpu_core" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "cpu_core@BR_INST_RETIRED.ALL_BRANCHES@ / cpu_core@B= R_INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch", @@ -1598,7 +1599,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW", + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW.", "Unit": "cpu_core" }, { @@ -1607,7 +1608,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1616,7 +1617,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1625,7 +1626,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1634,7 +1635,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1697,7 +1698,7 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 8 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 17", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp", "Unit": "cpu_core" }, @@ -1708,6 +1709,13 @@ "MetricName": "tma_info_memory_fb_hpki", "Unit": "cpu_core" }, + { + "BriefDescription": "Average per-thread data fill bandwidth to the= L1 data cache [GB / sec]", + "MetricExpr": "64 * cpu_core@L1D.REPLACEMENT@ / 1e9 / tma_info_sys= tem_time", + "MetricGroup": "Mem;MemoryBW", + "MetricName": "tma_info_memory_l1d_cache_fill_bw", + "Unit": "cpu_core" + }, { "BriefDescription": "Average per-thread data fill bandwidth to the= Level 0 within L1D cache [GB / sec]", "MetricExpr": "64 * cpu_core@L1D.L0_REPLACEMENT@ / 1e9 / tma_info_= system_time", @@ -1815,7 +1823,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp", "Unit": "cpu_core" @@ -1873,7 +1881,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand loads", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@mem_inst_retired.stlb_miss_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@MEM_INST_RETIRED.STLB_MISS_LOADS@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_load_stlb_miss_ret > 0.05", @@ -1896,7 +1904,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand stores", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@mem_inst_retired.stlb_miss_stores@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@MEM_INST_RETIRED.STLB_MISS_STORES@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_store_stlb_miss_ret > 0.05= ", @@ -1935,20 +1943,20 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@ASSISTS.ANY@", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)", "Unit": "cpu_core" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire", "Unit": "cpu_core" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1", @@ -1993,23 +2001,22 @@ }, { "BriefDescription": "Instructions per Far Branch ( Far Branches ap= ply upon transition from application to operating system, handling interrup= ts, exceptions) [lower number means higher occurrence rate]", - "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / BR_INST_RETIRED.FAR_BR= ANCH:u", + "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.FAR_BRANCH@u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000", + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6", "Unit": "cpu_core" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@INS= T_RETIRED.ANY_P@k", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / cpu_core@CPU_CLK_UNHA= LTED.THREAD@", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@CPU= _CLK_UNHALTED.THREAD@", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05", @@ -2053,7 +2060,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD@", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks", @@ -2064,7 +2071,6 @@ "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { @@ -2072,7 +2078,7 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_ISSU= ED.ANY@", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage", + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage.", "Unit": "cpu_core" }, { @@ -2084,7 +2090,7 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "cpu_core@TOPDOWN.SLOTS@", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots", "Unit": "cpu_core" @@ -2102,15 +2108,15 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@BR_= INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 8 * 1.5", + "MetricThreshold": "tma_info_thread_uptb < 12", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2120,7 +2126,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2129,8 +2135,8 @@ "MetricExpr": "cpu_core@INT_VEC_RETIRED.128BIT@ / (tma_retiring * = tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2139,8 +2145,8 @@ "MetricExpr": "cpu_core@INT_VEC_RETIRED.256BIT@ / (tma_retiring * = tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_ports_utilized_= 2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2149,8 +2155,8 @@ "MetricExpr": "cpu_core@ICACHE_TAG.STALLS@ / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2159,17 +2165,17 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L1@ / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit Level 1 after missing Level 0 within the L1D= cache", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * cpu_cor= e@mem_load_retired.l1_hit_l1@R, cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * 9) i= f 0 < cpu_core@mem_load_retired.l1_hit_l1@R else cpu_core@MEM_LOAD_RETIRED.= L1_HIT_L1@ * 9) / tma_info_thread_clks", + "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit Level 1 after missing Level 0 within the L1D= cache.", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * min(cpu_core= @MEM_LOAD_RETIRED.L1_HIT_L1@R, 9) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_capacity", - "MetricThreshold": "tma_l1_latency_capacity > 0.1 & tma_l1_bound >= 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_capacity > 0.1 & (tma_l1_bound = > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2178,8 +2184,8 @@ "MetricExpr": "4 * cpu_core@DEPENDENT_LOADS.ANY@ / tma_info_thread= _clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: DEPENDENT_LOADS.ANY", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2188,17 +2194,17 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L2@ / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles wit= h demand load accesses that hit the L2 cache under unloaded scenarios (poss= ibly L2 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * cpu_core@m= em_load_retired.l2_hit@R, cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info= _system_core_frequency)) if 0 < cpu_core@mem_load_retired.l2_hit@R else cpu= _core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info_system_core_frequency)) * (1= + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ /= 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L2_HIT@R, 3 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2208,18 +2214,18 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L3@ / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * cpu_core@m= em_load_retired.l3_hit@R, cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 * tma_inf= o_system_core_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_c= ore@mem_load_retired.l3_hit@R else cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 = * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequency) * (= 1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ = / 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L3_HIT@R, 9 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ranch_resteers, tma_mem_latency, tma_store_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2228,19 +2234,19 @@ "MetricExpr": "cpu_core@DECODE.LCP@ / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2250,7 +2256,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_load_op_utilization", "MetricThreshold": "tma_load_op_utilization > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Load operations. Sample with: = UOPS_DISPATCHED.LOAD", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Load operations. Sample with: = UOPS_DISPATCHED.PORT_2_3_10", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2259,7 +2265,7 @@ "MetricExpr": "max(0, tma_dtlb_load - tma_load_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2268,43 +2274,43 @@ "MetricExpr": "cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@ / tma_info_t= hread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_1G@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_2M_4M@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_C= OMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_4K@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU spent handling cache misses due to lock operations", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@me= m_inst_retired.lock_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@ME= M_INST_RETIRED.LOCK_LOADS@R / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2315,7 +2321,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2326,17 +2332,17 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_l1_bound= , tma_microcode_sequencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_sq_full", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2345,34 +2351,34 @@ "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD@) / tma_info_thread_clks - tm= a_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-mem\\-bound@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * cpu_core@MISC2_RETIRED.LFENCE@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * cpu_core@MEM_UOP_RETIRED.ANY= @ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -2395,14 +2401,14 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * cpu_= core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the MITE pipeline (the legacy deco= de pipeline)", - "MetricExpr": "(cpu_core@IDQ.MITE_UOPS\\,cmask\\=3D0x8\\,inv\\=3D0= x1@ / tma_info_thread_clks + cpu_core@IDQ.MITE_UOPS@ / (cpu_core@IDQ.DSB_UO= PS@ + cpu_core@IDQ.MITE_UOPS@) * (cpu_core@IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.= CORE@ - cpu_core@IDQ_BUBBLES.FETCH_LATENCY@)) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@IDQ.MITE_UOPS\\,cmask\\=3D0x8\\,inv\\=3D0= x1@ / 2 + cpu_core@IDQ.MITE_UOPS@ / (cpu_core@IDQ.DSB_UOPS@ + cpu_core@IDQ.= MITE_UOPS@) * (cpu_core@IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE@ - cpu_core@ID= Q_BUBBLES.FETCH_LATENCY@)) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", @@ -2411,17 +2417,17 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", "MetricExpr": "cpu_core@IDQ.MS_CYCLES_ANY@ / tma_info_thread_clks", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", @@ -2434,7 +2440,7 @@ "MetricExpr": "3 * cpu_core@IDQ.MS_SWITCHES@ / tma_info_thread_clk= s", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2445,7 +2451,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2454,7 +2460,7 @@ "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.NOP@ /= (tma_retiring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2470,20 +2476,20 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - cpu_core@BR_MISP_= RETIRED.ALL_BRANCHES@ / (cpu_core@INT_MISC.CLEARS_COUNT@ - cpu_core@MACHINE= _CLEARS.COUNT@)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - cpu_core@MACHINE_CLEA= RS.MEMORY_ORDERING@ / cpu_core@MACHINE_CLEARS.COUNT@), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2493,7 +2499,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2502,8 +2508,8 @@ "MetricExpr": "((cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ + (cpu_c= ore@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVITY.2_3_P= ORTS_UTIL@)) / tma_info_thread_clks if cpu_core@ARITH.DIV_ACTIVE@ < cpu_cor= e@CYCLE_ACTIVITY.STALLS_TOTAL@ - cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ else= (cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVIT= Y.2_3_PORTS_UTIL@) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2512,8 +2518,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ / tma_info= _thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2522,7 +2528,7 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2533,8 +2539,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.2_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma= _int_vector_256b", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2544,24 +2550,24 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.CYCLES_GE_3@ / tma_info_thre= ad_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@br_mi= sp_retired.ret_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@BR_MI= SP_RETIRED.RET_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ret_mispredicts", - "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispre= dicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & (tma_branch_mispr= edicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-retiring@ / (cpu_core@topdown\\-= fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiring@= + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2575,8 +2581,8 @@ "MetricExpr": "(cpu_core@BE_STALLS.SCOREBOARD@ + cpu_core@CPU_CLK_= UNHALTED.C02@) / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: BE_STALLS.SCOREBOARD. Related metrics: tm= a_ms_switches", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2585,8 +2591,8 @@ "MetricExpr": "tma_light_operations * cpu_core@INT_VEC_RETIRED.SHU= FFLES@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2596,28 +2602,28 @@ "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.PAUSE@ / tma_info_thread_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles hand= ling memory load split accesses - load that cross 64-byte cache line bounda= ry", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * cpu_c= ore@mem_inst_retired.split_loads@R, cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ = * tma_info_memory_load_miss_real_latency) if 0 < cpu_core@mem_inst_retired.= split_loads@R else cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * tma_info_memory= _load_miss_real_latency) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * min(cpu_co= re@MEM_INST_RETIRED.SPLIT_LOADS@R, tma_info_memory_load_miss_real_latency) = / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents rate of split store ac= cesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * cpu_= core@mem_inst_retired.split_stores@R, cpu_core@MEM_INST_RETIRED.SPLIT_STORE= S@) if 0 < cpu_core@mem_inst_retired.split_stores@R else cpu_core@MEM_INST_= RETIRED.SPLIT_STORES@) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * min(cpu_c= ore@MEM_INST_RETIRED.SPLIT_STORES@R, 1) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2626,8 +2632,8 @@ "MetricExpr": "(cpu_core@XQ.FULL@ + cpu_core@L1D_MISS.L2_STALLS@) = / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2636,8 +2642,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.BOUND_ON_STORES@ / tma_info_t= hread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2646,8 +2652,8 @@ "MetricExpr": "13 * cpu_core@LD_BLOCKS.STORE_FORWARD@ / tma_info_t= hread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2656,8 +2662,8 @@ "MetricExpr": "(cpu_core@MEM_STORE_RETIRED.L2_HIT@ * 10 * (1 - cpu= _core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.ALL_STORES@)= + (1 - cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.A= LL_STORES@) * min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFFCORE_REQUE= STS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO@)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2667,7 +2673,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_store_op_utilization", "MetricThreshold": "tma_store_op_utilization > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Store operations. Sample with:= UOPS_DISPATCHED.STD, UOPS_DISPATCHED.STA", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Store operations. Sample with:= UOPS_DISPATCHED.PORT_7_8", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2676,7 +2682,7 @@ "MetricExpr": "max(0, tma_dtlb_store - tma_store_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2685,34 +2691,34 @@ "MetricExpr": "cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@ / tma_info_= thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_1G@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_2M_4M@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_= core@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.W= ALK_COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_4K@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2721,7 +2727,7 @@ "MetricExpr": "9 * cpu_core@OCR.STREAMING_WR.ANY_RESPONSE@ / tma_i= nfo_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2731,7 +2737,7 @@ "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / tma_info= _thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2741,8 +2747,8 @@ "MetricExpr": "tma_retiring * cpu_core@UOPS_EXECUTED.X87@ / cpu_co= re@UOPS_EXECUTED.THREAD@", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%", "Unit": "cpu_core" } diff --git a/tools/perf/pmu-events/arch/x86/arrowlake/cache.json b/tools/pe= rf/pmu-events/arch/x86/arrowlake/cache.json index f63594b2cca8..f9ba410d4b94 100644 --- a/tools/perf/pmu-events/arch/x86/arrowlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/arrowlake/cache.json @@ -8,6 +8,16 @@ "SampleAfterValue": "1000003", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts the number of L1D cacheline (dirty) ev= ictions caused by load misses, stores, and prefetches.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x51", + "EventName": "DL1.DIRTY_EVICTION", + "PublicDescription": "Counts the number of L1D cacheline (dirty) e= victions caused by load misses, stores, and prefetches. Does not count evi= ctions or dirty writebacks caused by snoops. Does not count a replacement = unless a (dirty) line was written back.", + "SampleAfterValue": "200003", + "UMask": "0x1", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Counts the number of cache lines replaced in = L0 data cache.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -18,6 +28,16 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Cachelines replaced into the L0 and L1 d-cach= e. Successful replacements only (not blocked) and exclude WB-miss case", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x51", + "EventName": "L1D.REPLACEMENT", + "PublicDescription": "Counts cachelines replaced into the L0 and L= 1 d-cache.", + "SampleAfterValue": "1000003", + "UMask": "0x5", + "Unit": "cpu_core" + }, { "BriefDescription": "Number of cycles a demand request has waited = due to L1D Fill Buffer (FB) unavailability.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -79,6 +99,46 @@ "UMask": "0x1f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Exclusive state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.E", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Exclusive state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Forward state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.F", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Forward state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x10", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Modified state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.M", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Modified state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Shared state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.S", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Shared state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Modified cache lines that are evicted by L2 c= ache when triggered by an L2 cache fill.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -89,6 +149,16 @@ "UMask": "0x2", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = evicted due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.NON_SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= evicted due to an L2 cache fill. Increments on the core that brought the l= ine in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Non-modified cache lines that are silently dr= opped by L2 cache.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -99,6 +169,16 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = silently dropped due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= silently dropped due to an L2 cache fill. Increments on the core that bro= ught the line in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Cache lines that have been L2 hardware prefet= ched but not used by demand accesses", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -128,6 +208,15 @@ "UMask": "0xff", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that r= esulted in a Hit from a front door request only (does not include rejects o= r recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.HIT", + "SampleAfterValue": "200003", + "UMask": "0x2", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Read requests with true-miss in L2 cache [Thi= s event is alias to L2_RQSTS.MISS]", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -138,6 +227,34 @@ "UMask": "0x3f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of total L2 Cache Accesses = that resulted in a Miss from a front door request only (does not include re= jects or recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.MISS", + "SampleAfterValue": "200003", + "UMask": "0x1", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that m= iss the L2 and get BBL reject short and long rejects (includes those count= ed in L2_reject_XQ.any), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.REJECTS", + "SampleAfterValue": "200003", + "UMask": "0x4", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "L2 code requests", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x24", + "EventName": "L2_RQSTS.ALL_CODE_RD", + "PublicDescription": "Counts the total number of L2 code requests.= ", + "SampleAfterValue": "200003", + "UMask": "0xe4", + "Unit": "cpu_core" + }, { "BriefDescription": "Demand Data Read access L2 cache", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -408,6 +525,15 @@ "UMask": "0x78", "Unit": "cpu_lowpower" }, + { + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled to a store buffer full condition", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x34", + "EventName": "MEM_BOUND_STALLS_LOAD.SBFULL", + "SampleAfterValue": "1000003", + "UMask": "0x80", + "Unit": "cpu_lowpower" + }, { "BriefDescription": "Counts all retired load instructions.", "Counter": "0,1,2,3", @@ -1245,6 +1371,17 @@ "UMask": "0xf", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache where a snoop hit in another cores caches, data forwarding i= s required as the data is modified.", "Counter": "0,1,2,3", @@ -1267,6 +1404,17 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that were su= pplied by the L3 cache where a snoop hit in another cores caches, data forw= arding is required as the data is modified.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/arrowlake/memory.json b/tools/p= erf/pmu-events/arch/x86/arrowlake/memory.json index 08f01fc66fef..f7e202dec84a 100644 --- a/tools/perf/pmu-events/arch/x86/arrowlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/arrowlake/memory.json @@ -332,6 +332,17 @@ "UMask": "0x4", "Unit": "cpu_lowpower" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1E780000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/arrowlake/other.json b/tools/pe= rf/pmu-events/arch/x86/arrowlake/other.json index 0175b2193201..f4fb51bb95ff 100644 --- a/tools/perf/pmu-events/arch/x86/arrowlake/other.json +++ b/tools/perf/pmu-events/arch/x86/arrowlake/other.json @@ -18,71 +18,6 @@ "UMask": "0x8", "Unit": "cpu_core" }, - { - "BriefDescription": "Counts cycles where the pipeline is stalled d= ue to serializing operations.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa2", - "EventName": "BE_STALLS.SCOREBOARD", - "SampleAfterValue": "100003", - "UMask": "0x2", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Count number of times a load is depending on = another load that had just write back its data or in previous or 2 cycles = back. This event supports in-direct dependency through a single uop.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x02", - "EventName": "DEPENDENT_LOADS.ANY", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts the number of uops executed on seconda= ry integer ports 0,1,2,3.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.2ND", - "SampleAfterValue": "1000003", - "UMask": "0x80", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on a load = port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.LD", - "PublicDescription": "Counts the number of uops executed on a load= port. This event counts for integer uops even if the destination is FP/ve= ctor", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 0,1, 2, 3.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.PRIMARY", - "SampleAfterValue": "1000003", - "UMask": "0x78", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on a Store= address port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.STA", - "PublicDescription": "Counts the number of uops executed on a Stor= e address port. This event counts integer uops even if the data source is F= P/vector", - "SampleAfterValue": "1000003", - "UMask": "0x2", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on an inte= ger store data and jump port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.STD_JMP", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, { "BriefDescription": "This event is deprecated. [This event is alia= s to MISC_RETIRED.LBR_INSERTS]", "Counter": "0,1,2,3,4,5,6,7", @@ -93,75 +28,6 @@ "UMask": "0x1", "Unit": "cpu_lowpower" }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L1 cache (that is: no execution & load in flight = & no load missed L1 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L1", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L2 cache (that is: no execution & load in flight = & load missed L1 & no load missed L2 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L2", - "SampleAfterValue": "1000003", - "UMask": "0x2", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L3 cache (that is: no execution & load in flight = & load missed L1 & load missed L2 cache & no load missed L3 Cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L3", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for Memory (that is: no execution & load in flight & = a load missed L3 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.MEM", - "SampleAfterValue": "1000003", - "UMask": "0x8", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1E780000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3,4,5,6,7", @@ -206,65 +72,6 @@ "UMask": "0x1", "Unit": "cpu_core" }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4", - "Unit": "cpu_lowpower" - }, - { - "BriefDescription": "Counts the number of issue slots where no uop= could issue due to an IQ scoreboard that stalls allocation until a specifi= ed older uop retires or (in the case of jump scoreboard) executes. Commonly= executed instructions with IQ scoreboards include LFENCE and MFENCE.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.IQ_JEU_SCB", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3,4,5,6,7,8,9", diff --git a/tools/perf/pmu-events/arch/x86/arrowlake/pipeline.json b/tools= /perf/pmu-events/arch/x86/arrowlake/pipeline.json index 6dbde51e7ead..739efb199668 100644 --- a/tools/perf/pmu-events/arch/x86/arrowlake/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/arrowlake/pipeline.json @@ -51,6 +51,15 @@ "UMask": "0x1f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts cycles where the pipeline is stalled d= ue to serializing operations.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa2", + "EventName": "BE_STALLS.SCOREBOARD", + "SampleAfterValue": "100003", + "UMask": "0x2", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts the total number of branch instruction= s retired for all branch types.", "Counter": "0,1,2,3,4,5,6,7", @@ -888,6 +897,15 @@ "UMask": "0x4", "Unit": "cpu_core" }, + { + "BriefDescription": "Count number of times a load is depending on = another load that had just write back its data or in previous or 2 cycles = back. This event supports in-direct dependency through a single uop.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x02", + "EventName": "DEPENDENT_LOADS.ANY", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, { "BriefDescription": "Cycles total of 1 uop is executed on all port= s and Reservation Station was not empty.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1139,6 +1157,53 @@ "UMask": "0x10", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of uops executed on seconda= ry integer ports 0,1,2,3.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.2ND", + "SampleAfterValue": "1000003", + "UMask": "0x80", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on a load = port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.LD", + "PublicDescription": "Counts the number of uops executed on a load= port. This event counts for integer uops even if the destination is FP/ve= ctor", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 0,1, 2, 3.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.PRIMARY", + "SampleAfterValue": "1000003", + "UMask": "0x78", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on a Store= address port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.STA", + "PublicDescription": "Counts the number of uops executed on a Stor= e address port. This event counts integer uops even if the data source is F= P/vector", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on an inte= ger store data and jump port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.STD_JMP", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, { "BriefDescription": "Number of vector integer instructions retired= of 128-bit vector-width.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1405,8 +1470,9 @@ "Unit": "cpu_atom" }, { - "BriefDescription": "Counts the number of machine clears that flus= h the pipeline and restart the machine with the use of microcode due to SMC= , MEMORY_ORDERING, FP_ASSISTS, PAGE_FAULT, DISAMBIGUATION, and FPC_VIRTUAL_= TRAP.", + "BriefDescription": "This event is deprecated.", "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", "EventCode": "0xc3", "EventName": "MACHINE_CLEARS.SLOW", "SampleAfterValue": "20003", @@ -1432,6 +1498,42 @@ "UMask": "0x1", "Unit": "cpu_lowpower" }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L1 cache (that is: no execution & load in flight = & no load missed L1 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L1", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L2 cache (that is: no execution & load in flight = & load missed L1 & no load missed L2 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L2", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L3 cache (that is: no execution & load in flight = & load missed L1 & load missed L2 cache & no load missed L3 Cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L3", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for Memory (that is: no execution & load in flight & = a load missed L3 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.MEM", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_core" + }, { "BriefDescription": "LFENCE instructions retired", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1460,6 +1562,65 @@ "UMask": "0x1", "Unit": "cpu_lowpower" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4", + "Unit": "cpu_lowpower" + }, + { + "BriefDescription": "Counts the number of issue slots where no uop= could issue due to an IQ scoreboard that stalls allocation until a specifi= ed older uop retires or (in the case of jump scoreboard) executes. Commonly= executed instructions with IQ scoreboards include LFENCE and MFENCE.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.IQ_JEU_SCB", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts the number of issue slots not consumed= by the backend due to a micro-sequencer (MS) scoreboard, which stalls the = front-end from issuing from the UROM until a specified older uop retires.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 0ef31b65f8df..1b592cf63940 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -1,7 +1,7 @@ Family-model,Version,Filename,EventType GenuineIntel-6-(97|9A|B7|BA|BF),v1.29,alderlake,core GenuineIntel-6-BE,v1.29,alderlaken,core -GenuineIntel-6-C[56],v1.07,arrowlake,core +GenuineIntel-6-C[56],v1.08,arrowlake,core GenuineIntel-6-(1C|26|27|35|36),v5,bonnell,core GenuineIntel-6-(3D|47),v30,broadwell,core GenuineIntel-6-56,v12,broadwellde,core --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B3D8D2C9D for ; Wed, 26 Mar 2025 00:05:39 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947541; cv=none; b=Cu0/q37XJlge4+lVqEXyYC8d41cKtU26pV+hnWm6oUjxpXzmINy5xtQhbgMvf/sD7ffoGgc6RpYVQRmfRp6kPyYExXsxQhxPcN9pydIL4EYfrONtM6KIUc/MBZJlRJWKvavq/lXCF+UwvZlCOaqtwslV348kqX6KCMPZMdCbY60= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947541; c=relaxed/simple; bh=gzOhUwuk6PRtqJT004lFT8JmHmf233ItpyrfAz7UByY=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=gNCs/7noFcfTK87rD/59F8VnAVfOKFyb71VghyEZOpomdD8Zo28SCm/tojGdmb5y9+0Zar1VZ2nASMBhidA3+wCPiHoft4QLoU6VwQ54jxpjzXnjeolUuJWkl90bv61QkK/4ppFtdvi/VDvDGaTUKtiQfPb+GlMep23uzWaLMho= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=QIBybslv; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="QIBybslv" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff55176edcso10558741a91.1 for ; Tue, 25 Mar 2025 17:05:39 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947539; x=1743552339; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=6RXv0a2X2Eq8/LheD1NmgszijzevuXG/6uAHQeKi4nU=; b=QIBybslvucyY+IHEtWVssnwuCH/gYZmf91/146mcCNUdz0iJUwcjvqkmtMVhZikhec FcLUfmDOlvJQ9+hxW6CBueR5Fdm5AvwGFdL1iCF9puZGaNYqefibBND8AXPUC40FlXmH jECc44vXq34iNWcrm/+eUGish9eV6ksFFfcH9bblPdWSME14RL/lhqy0HvPK455VhKZd hUlNtOL8C8HbRhSJ5n6T0b8lDiGl3WWVOt2TOzVRWyByWsIU6WNPxgVerSPvD26BS4RD fYNg3BTR7x9TPJ7tY93/Nt2LjXGBglvBq6XLAfs7QH25NJLT8gCAUq5mfHu0ZeaF5znr uUaA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947539; x=1743552339; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=6RXv0a2X2Eq8/LheD1NmgszijzevuXG/6uAHQeKi4nU=; b=TcfaSO+Pxb3TAbAHVN19GiOmOdsw1V4Gb+tfIuL7n0tLI+S+ad7+IycSmMbrgPd0Ks iPy4kNMmkWCBWCg8WF8qzkyeZZyVLZd0cyLuIAnvTvH0mnQTvi8ZM6ACm3tIlT01YtRK J1+yA7gxw1ggpJZ+NKFnM9/vsMhqZS89an1fQ74O9/JS4QhpPZ3pzg6zJCq/lHowzLyz LouIm3AQ8clF4/JKXtERrLzwW4sImSU1KKiuWAkRK11xPW+qso2MkCVqX6U/dElmUW0z DAEz0yJJkaPuJ+SgYjdHUGx6DPjKrIaVlvG1+0/z+wCbtwuGejYufbvcEADzqT98XSdl G7Gg== X-Forwarded-Encrypted: i=1; AJvYcCWYQLGvV1d9s58K2Ez1lm07yKlC2A8hEsAjxAk0jaiDoBQDFBE7v4xFKZRODm3k/BmwZRu7sKrMqziWIek=@vger.kernel.org X-Gm-Message-State: AOJu0YwOl7RRdfGkkIz/PWrMAJsgRdoTDh9m1nFH+8kaWogtFGgPLAzP rqC+nRO7x7tpqwm95u9T29PitbQMcI77bhgyhRs0uv2vBQNRXwxSWnd99oajqF2MZVqVdKUhQmG Pm/zCfw== X-Google-Smtp-Source: AGHT+IE4oWDbWFvtLNFUwOWfpQXhRUeiwXDJOaV/ZyS/d/4GQ1851TdG55OmtLkNTLtGGaToqQ3bXUo2qjqi X-Received: from pgbdo14.prod.google.com ([2002:a05:6a02:e8e:b0:af2:48c3:360f]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:688:b0:1f5:7353:c303 with SMTP id adf61e73a8af0-1fe42f2cbf3mr30460960637.11.1742947539131; Tue, 25 Mar 2025 17:05:39 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:02 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-5-irogers@google.com> Subject: [PATCH v3 04/35] perf vendor events: Update bonnell events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Move DISPATCH_BLOCKED.ANY to the pipeline topic. Signed-off-by: Ian Rogers --- tools/perf/pmu-events/arch/x86/bonnell/other.json | 8 -------- tools/perf/pmu-events/arch/x86/bonnell/pipeline.json | 8 ++++++++ 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/bonnell/other.json b/tools/perf= /pmu-events/arch/x86/bonnell/other.json index 3a55c101fbf7..6e6f64b96834 100644 --- a/tools/perf/pmu-events/arch/x86/bonnell/other.json +++ b/tools/perf/pmu-events/arch/x86/bonnell/other.json @@ -323,14 +323,6 @@ "SampleAfterValue": "2000000", "UMask": "0x2" }, - { - "BriefDescription": "Memory cluster signals to block micro-op disp= atch for any reason", - "Counter": "0,1", - "EventCode": "0x9", - "EventName": "DISPATCH_BLOCKED.ANY", - "SampleAfterValue": "200000", - "UMask": "0x20" - }, { "BriefDescription": "Number of Enhanced Intel SpeedStep(R) Technol= ogy (EIST) transitions", "Counter": "0,1", diff --git a/tools/perf/pmu-events/arch/x86/bonnell/pipeline.json b/tools/p= erf/pmu-events/arch/x86/bonnell/pipeline.json index 9ff032ab11e2..48d3d053a369 100644 --- a/tools/perf/pmu-events/arch/x86/bonnell/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/bonnell/pipeline.json @@ -211,6 +211,14 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, + { + "BriefDescription": "Memory cluster signals to block micro-op disp= atch for any reason", + "Counter": "0,1", + "EventCode": "0x9", + "EventName": "DISPATCH_BLOCKED.ANY", + "SampleAfterValue": "200000", + "UMask": "0x20" + }, { "BriefDescription": "Divide operations retired", "Counter": "0,1", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oo1-f73.google.com (mail-oo1-f73.google.com [209.85.161.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 8DF85E56F for ; Wed, 26 Mar 2025 00:05:48 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.161.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947554; cv=none; b=Y0eby3PSIXS5Km7Y6NMG1aOlh1uvCCOGL98JzpXfii/qVGepdm7K18vRCD6KmFLb89dVVdmKHVwtc+zYJDO3K31ksWC3WEfKOu+juwaDLgg8bGerMKDdBQmRE5h2JZsXevz9EcqCvM6v5urGEaFOfsVC9F38IDOr3g+viOwzg80= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947554; c=relaxed/simple; bh=U6fdboGKnD4AJzFkvDZ3+eZqfEtKEmb9evjgAcZCmj4=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=SNykdcIVNh+2SfzBdJyonBqumEbgSNW6KN2YCgoGQFdcnIT3TpADZ82lphvC4149BSGJgxCoJsBt23ADd9YzFDycE4oijR/7pYNsb8Dogh8UtifJxU+t8prDHD9aXjxQjQHy8C/u/iGhSkEuQj9eQqBGYzj3NvscMn4irMu6xuQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=UWFr43am; arc=none smtp.client-ip=209.85.161.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="UWFr43am" Received: by mail-oo1-f73.google.com with SMTP id 006d021491bc7-6022020dd2bso1430410eaf.2 for ; Tue, 25 Mar 2025 17:05:48 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947547; x=1743552347; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=a5jocNXaBG8YnVIsRZJDaZAuXaovCUJdQml+Wnx7Beg=; b=UWFr43am0Rn3WguA+D0fNlP9ih2izdAZa0SBhk8AGAsO8gEAZxYWQYeIFlw4DkGjpc 9HUDevNbrKiSTCDQPTlvsH7EdOop2LJGxhNBAma3JwvEgD6xqiOPtkvbdQZJ9lsyxkzW tcKaGToUEIVR4Bntl5S0178HvjMZI5OWUCFBn42LafOt+9vAkRxIlvCrX/15ZLRx0Nbj F9kaGQhTzsklu4vjYowSY0Gb5F+9/nbSCz1J1GycNU3MDHOjz+1ynOr1ikRxEmmh5CX0 yNhXm6frPPMZsZ5qyGwlWSVV4/ZuNRJ5Y49xuL5hIrbLRMcrJuD0JBgxx5JvjfsjxR+y WhdA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947547; x=1743552347; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=a5jocNXaBG8YnVIsRZJDaZAuXaovCUJdQml+Wnx7Beg=; b=NAqGR3F6Yf3t56mRTWDFOeRnMM2cHEM8eaTiAhXSxIyVRWl0WPtM11jDqQIxWYxHWj L1JBkgiNgxwZa46/olAHaE6Oh8+D3Uo/Sn0I49V6zaHa0ZIH8svWL744u2hXYNfd1BFm E7LhDh0X4636mch8ziqpdllm3b/lwW3D2TshjuK9nS3h6FyOGC3nXHGT2fhLa6XPxsxX rPzEuYLkF5oiCO+BrOaRWqfFtB0jRkh0bND6u0SFdC5trt40o+hQShs1wtO9vJVGPyrX r3EA7SFiwAcxpa1Ms88TV+vovj6XMqsZuGNTUtc3K4Qq7lWWC//ptsUTN694oLGN865B vUDA== X-Forwarded-Encrypted: i=1; AJvYcCUhXIwEaRx5hKzOnjF2eM5oLExKEzfFo7WZak+mPQB7gviEW5uSmCZsERnWuMBl53ojjo6MbzNi4PKWvaA=@vger.kernel.org X-Gm-Message-State: AOJu0YymnYrIeyMWi+Blh7HAeemQnR+9SrEUs/jza1D/hvi7jxCjFD/P QDBFMjb28St1sIZFmrVdgMv2p0ye3ecX9DcJeBhDiOPFuoQppTI7DQaCS3od7clp4ehBnHPR4nY kK7E17Q== X-Google-Smtp-Source: AGHT+IEOzoELfuz91FMMKCIH/I6uC99wg+uCbjChVVtVJsvppnn0H6hP5qNXdHKRwD4kfLEzGXf2l+32QRTB X-Received: from oachu20.prod.google.com ([2002:a05:6871:5294:b0:2b8:431f:3a1d]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6870:569e:b0:2c2:c92a:5781 with SMTP id 586e51a60fabf-2c7804e8d05mr13093889fac.27.1742947547393; Tue, 25 Mar 2025 17:05:47 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:03 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-6-irogers@google.com> Subject: [PATCH v3 05/35] perf vendor events: Update broadwell metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/broadwell/bdw-metrics.json | 256 +++++++++--------- 1 file changed, 127 insertions(+), 129 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/broadwell/bdw-metrics.json b/to= ols/perf/pmu-events/arch/x86/broadwell/bdw-metrics.json index 40970fa5566c..89750117a7f6 100644 --- a/tools/perf/pmu-events/arch/x86/broadwell/bdw-metrics.json +++ b/tools/perf/pmu-events/arch/x86/broadwell/bdw-metrics.json @@ -74,12 +74,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", @@ -92,8 +92,8 @@ "MetricExpr": "66 * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_= slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y_WB_ASSIST", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, { @@ -104,7 +104,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -114,7 +114,7 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { @@ -125,7 +125,7 @@ "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_mispredicts_resteers", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers", "ScaleUnit": "100%" }, { @@ -133,8 +133,8 @@ "MetricExpr": "12 * (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS= .COUNT + BACLEARS.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -143,8 +143,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -152,7 +152,7 @@ "MetricExpr": "MACHINE_CLEARS.COUNT * tma_branch_resteers / (BR_MI= SP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Rel= ated metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches", "ScaleUnit": "100%" }, @@ -162,8 +162,8 @@ "MetricExpr": "(60 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * (1 = + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_= UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS= _L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LO= AD_UOPS_RETIRED.L3_MISS))) + 43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS *= (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_L= OAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + ME= M_LOAD_UOPS_RETIRED.L3_MISS)))) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MIS= S. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears= ", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { @@ -174,7 +174,7 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { @@ -183,8 +183,8 @@ "MetricExpr": "43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * (1 + = MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UO= PS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L= 3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD= _UOPS_RETIRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, t= ma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_contested_accesses, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -192,8 +192,8 @@ "MetricExpr": "ARITH.FPU_DIV_ACTIVE / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.FPU_DIV_ACTIVE", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -202,8 +202,8 @@ "MetricExpr": "(1 - MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_= RETIRED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS)) * CYCLE_ACTIVITY.STALL= S_L2_MISS / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS_PS", "ScaleUnit": "100%" }, { @@ -212,7 +212,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -220,26 +220,26 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Related metrics: tma_fetch_bandw= idth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISS= ES.WALK_DURATION\\,cmask\\=3D0x1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED) / = tma_info_thread_clks", + "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISS= ES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED) / tm= a_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MI= SSES.WALK_DURATION\\,cmask\\=3D0x1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED)= / tma_info_thread_clks", + "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MI= SSES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED) /= tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -247,18 +247,18 @@ "MetricExpr": "60 * OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM = / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3= _HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Rela= ted metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -287,7 +287,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -295,8 +295,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports= _utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -304,8 +304,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -313,8 +313,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -322,8 +322,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -333,33 +333,33 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound.", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses.", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -370,7 +370,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -391,11 +391,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -420,7 +420,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -438,7 +438,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -446,7 +446,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -454,7 +454,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -462,7 +462,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -470,7 +470,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -512,7 +512,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, = tma_lcp" }, { @@ -634,20 +634,20 @@ }, { "BriefDescription": "Utilization of the core's Page Walker(s) serv= ing STLB misses triggered by instruction/Load/Store accesses", - "MetricExpr": "(cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=3D0x1@ + c= pu@DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=3D0x1@ + cpu@DTLB_STORE_MISSES.= WALK_DURATION\\,cmask\\=3D0x1@ + 7 * (DTLB_STORE_MISSES.WALK_COMPLETED + DT= LB_LOAD_MISSES.WALK_COMPLETED + ITLB_MISSES.WALK_COMPLETED)) / tma_info_cor= e_core_clks", + "MetricExpr": "(cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=3D1@ + cpu= @DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=3D1@ + cpu@DTLB_STORE_MISSES.WALK= _DURATION\\,cmask\\=3D1@ + 7 * (DTLB_STORE_MISSES.WALK_COMPLETED + DTLB_LOA= D_MISSES.WALK_COMPLETED + ITLB_MISSES.WALK_COMPLETED)) / tma_info_core_core= _clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_page_walks_utilization", "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D0x1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -688,14 +688,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -743,7 +742,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -752,15 +751,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -786,14 +784,14 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", - "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_D= URATION\\,cmask\\=3D0x1@ + 7 * ITLB_MISSES.WALK_COMPLETED) / tma_info_threa= d_clks", + "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_D= URATION\\,cmask\\=3D1@ + 7 * ITLB_MISSES.WALK_COMPLETED) / tma_info_thread_= clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_M= ISSES.WALK_COMPLETED", "ScaleUnit": "100%" }, @@ -802,8 +800,8 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metri= cs: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_m= s_switches, tma_ports_utilized_1", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -811,8 +809,8 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.ST= ALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT_PS", "ScaleUnit": "100%" }, { @@ -821,8 +819,8 @@ "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_RETIR= ED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS) * CYCLE_ACTIVITY.STALLS_L2_M= ISS / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { @@ -831,8 +829,8 @@ "MetricExpr": "29 * (MEM_LOAD_UOPS_RETIRED.L3_HIT * (1 + MEM_LOAD_= UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRE= D.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RET= IRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_UOPS_RET= IRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: = tma_branch_resteers, tma_mem_latency, tma_store_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS. Related metric= s: tma_mem_latency", "ScaleUnit": "100%" }, { @@ -840,18 +838,18 @@ "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage,= tma_info_inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -870,8 +868,8 @@ "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, { @@ -882,15 +880,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sh= aring, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sh= aring, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_c= ache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", "ScaleUnit": "100%" }, @@ -899,7 +897,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", "ScaleUnit": "100%" }, @@ -911,7 +909,7 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { @@ -928,8 +926,8 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES * tma_branch_resteers = / (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Related metrics: tma_branch_mispredicts", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Related metrics: tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, { @@ -938,7 +936,7 @@ "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", "ScaleUnit": "100%" }, { @@ -946,8 +944,8 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mix= ing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, { @@ -956,7 +954,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, = tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -965,7 +963,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_= utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1001,7 +999,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_sc= alar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tm= a_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1010,7 +1008,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, t= ma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1028,43 +1026,43 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES= _GE_1_UOP_EXEC - (UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if tma_info_thread_ip= c > 1.8 else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) - (RS_EVENTS.EMPTY_CYCLES= if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB - RESOURCE_STALLS.= SB - CYCLE_ACTIVITY.STALLS_MEM_ANY) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed no uops on any execution port (Logical Processor cycles since ICL= , Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\=3D0x1\\,cmask\\=3D0= x1@ / 2 if #SMT_on else CYCLE_ACTIVITY.STALLS_TOTAL - (RS_EVENTS.EMPTY_CYCL= ES if tma_fetch_latency > 0.1 else 0)) / tma_info_core_core_clks", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\,cmask\\=3D1@ / 2 if= #SMT_on else (CYCLE_ACTIVITY.STALLS_TOTAL - (RS_EVENTS.EMPTY_CYCLES if tma= _fetch_latency > 0.1 else 0)) / tma_info_core_core_clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re the CPU executed total of 1 uop per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x2@) / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_= GE_1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) / tma_info_core_core_c= lks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D2@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) / tma_info_core_core_clks= )", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 2 uops per cycle on all execution ports (Logical Process= or cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x3@) / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_= GE_2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_= clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D3@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clk= s)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ / 2 if #SM= T_on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clks", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ / 2 if #SMT_= on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1084,7 +1082,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1092,8 +1090,8 @@ "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_c= ore_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1101,7 +1099,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", "ScaleUnit": "100%" }, @@ -1110,8 +1108,8 @@ "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1119,8 +1117,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1129,8 +1127,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1146,7 +1144,7 @@ "MetricExpr": "tma_branch_resteers - tma_mispredicts_resteers - tm= a_clears_resteers", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1155,8 +1153,8 @@ "MetricExpr": "INST_RETIRED.X87 * tma_info_thread_uoppi / UOPS_RET= IRED.RETIRE_SLOTS", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" } ] --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 31FCF10E4 for ; Wed, 26 Mar 2025 00:05:57 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947561; cv=none; b=WiKDI9R0H7b080HyiHO+pMJBRywkM546F32FahejNmlp2POPoLXzZCe4nfWwVWFVqpz4N/5VQuuyzuhoMSkph4rV9YhDSHeZPiym1i1ck4vKlvp0TCe3Wnlc1E9JLQ1zLZgIfLC6T5G+O3f9+rdEmXQiUUbfaRbBIA0SuDQHRQM= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947561; c=relaxed/simple; bh=qJ/gTxANpq7Ae1pZopFMcULf+/ebK0e0kNg5YFqePCc=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=rezwVN0ATACo7zPRECkBnChHqCDxEIIt+SDxPGzDU8DN6egbKEIIMcDLXSM8QtICPrivPBJbE62Bubm2/TKS8LF4QluMZUN6Qg4PsZLhw5Ml46YvaJEz8Dxi2/5AlRMi0Ieh6Q7OS9RmwILN1peEQzXfZ/0oxqnj2XRZTD2v7TE= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=XSjnaKqX; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="XSjnaKqX" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2242ade807fso174328155ad.2 for ; Tue, 25 Mar 2025 17:05:57 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947557; x=1743552357; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=vdDcHKM0jXJHo/S/f6j1ZXtaUryEcuHG2tCg7U7Ajnk=; b=XSjnaKqXaJNlkzOjxBRlRsl4wT60gy6nAQqxJstadZ5yhRnr5qnYFAfQCvv20EC1E2 1XpbB7q3G/yQz44aXu+7LeF81Okv+ApfxYcKKS2vkWRE9QdCWlr0pRq3Qcfyp/PBuaqF CKeIbV/T/6PRjbEdxwXj4k60Zl5+ks6q9MnnZUWeY5kI6Xip6TTUPfnrL99s0iQlYoMb 3BSt+N8mM0Md1qpNRf+xIU1snGgeo/HGGq7Xj9IuQKXIZUH+IgjBe/UPmPkq6UAPxbx8 TqL01a6qLf29tBg3M3yWVixSYNmbKdRNLRPFC+NbLdKH/INEPW3sdOCAuiEpb7LZAmFN OKjQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947557; x=1743552357; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=vdDcHKM0jXJHo/S/f6j1ZXtaUryEcuHG2tCg7U7Ajnk=; b=gjDkxX4btC7R8934Q2FiTSY7i6cTxad7CdST6cnBqWmuTtak5K34/uS8Qx81mYklL8 0HZDQkWcmZUAG3xcgdc+aC3dTOISCCaq9yGoc9sjzS5HkgQOn5SIMMD5TG0XszDbTIg/ tbemlGJFAIKKHkx4wZkJeACBjevn8qG5UUWX8Cel3tCX3tw4iLzK9Geoef8PpJ3DoI+Z DwYq7v6jh6XVIWqw9ZpGS1UjeKE7Y33WuT+rBgZJG9y2ZfZvqE0QnPJ1PR1XrY+v+gxz gXdwzUt01Z9yu2NHQeNrfLxhLDSOAyrKGdgHh0CrZa3thVT17y26PySQnEV08NyWU7yO 6U9Q== X-Forwarded-Encrypted: i=1; AJvYcCUcJkhehr56ywg6k2lyrz5M20S6xk3qgztq7oVudNpOB5UyvU5RA4sk1Z2UI/7QbspZIwrWepZlJU5Pg9E=@vger.kernel.org X-Gm-Message-State: AOJu0YyJ2ZQJ8HvD8uoRu8My+81DSy3RoZ/Nl+2FGJ47bskaW7FRHwlC RvK2hNKnqrmvS9xqI2RnpLDdZx3Eq2yewWVwgmky951lwwCaDWBMmE9uoWXIi34beuSSSXOfex8 uQTBH3Q== X-Google-Smtp-Source: AGHT+IFyvKq/FouaLXfdQetuisMx/Cp0k7t+3IuR5KXCx586TzxJ7lZCX2GNwVUHCPnty/tl8gAa4BgFVddu X-Received: from pfnz17.prod.google.com ([2002:aa7:85d1:0:b0:736:451f:b9f4]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:1bc4:b0:736:5dc6:a14f with SMTP id d2e1a72fcca58-73905a35abbmr30018034b3a.23.1742947556684; Tue, 25 Mar 2025 17:05:56 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:04 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-7-irogers@google.com> Subject: [PATCH v3 06/35] perf vendor events: Update broadwellde metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/broadwellde/bdwde-metrics.json | 180 +++++++++--------- 1 file changed, 90 insertions(+), 90 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/broadwellde/bdwde-metrics.json = b/tools/perf/pmu-events/arch/x86/broadwellde/bdwde-metrics.json index b03a5f2bcd82..81175f0f2603 100644 --- a/tools/perf/pmu-events/arch/x86/broadwellde/bdwde-metrics.json +++ b/tools/perf/pmu-events/arch/x86/broadwellde/bdwde-metrics.json @@ -74,7 +74,7 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "(tma_4k_aliasing > 0.2) & ((tma_l1_bound > 0.1= ) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, @@ -84,7 +84,7 @@ "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", - "MetricThreshold": "(tma_alu_op_utilization > 0.4)", + "MetricThreshold": "tma_alu_op_utilization > 0.4", "ScaleUnit": "100%" }, { @@ -92,7 +92,7 @@ "MetricExpr": "66 * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_= slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "(tma_assists > 0.1) & ((tma_microcode_sequence= r > 0.05) & ((tma_heavy_operations > 0.1)))", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, @@ -102,7 +102,7 @@ "MetricExpr": "1 - (tma_frontend_bound + tma_bad_speculation + tma= _retiring)", "MetricGroup": "BvOB;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound > 0.2)", + "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS", "ScaleUnit": "100%" @@ -112,7 +112,7 @@ "MetricExpr": "(UOPS_ISSUED.ANY - UOPS_RETIRED.RETIRE_SLOTS + 4 * = (INT_MISC.RECOVERY_CYCLES_ANY / 2 if #SMT_on else INT_MISC.RECOVERY_CYCLES)= ) / tma_info_thread_slots", "MetricGroup": "TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation > 0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" @@ -123,7 +123,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * tma_bad_speculation", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts > 0.1) & ((tma_bad_spe= culation > 0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: = tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers", "ScaleUnit": "100%" @@ -133,7 +133,7 @@ "MetricExpr": "12 * (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS= .COUNT + BACLEARS.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "(tma_branch_resteers > 0.05) & ((tma_fetch_lat= ency > 0.1) & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, @@ -143,7 +143,7 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc > 0.1) & ((tma_microcode_sequencer >= 0.05) & ((tma_heavy_operations > 0.1)))", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, @@ -152,7 +152,7 @@ "MetricExpr": "MACHINE_CLEARS.COUNT * tma_branch_resteers / (BR_MI= SP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "(tma_clears_resteers > 0.05) & ((tma_branch_re= steers > 0.05) & ((tma_fetch_latency > 0.1) & ((tma_frontend_bound > 0.15))= ))", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, @@ -162,7 +162,7 @@ "MetricExpr": "(60 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * (1 = + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_= UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS= _L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LO= AD_UOPS_RETIRED.L3_MISS))) + 43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS *= (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_L= OAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + ME= M_LOAD_UOPS_RETIRED.L3_MISS)))) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "(tma_contested_accesses > 0.05) & ((tma_l3_bou= nd > 0.05) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma_remote= _cache", "ScaleUnit": "100%" }, @@ -172,7 +172,7 @@ "MetricExpr": "tma_backend_bound - tma_memory_bound", "MetricGroup": "Backend;Compute;TmaL2;TopdownL2;tma_L2_group;tma_b= ackend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound > 0.1) & ((tma_backend_bound >= 0.2))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" @@ -183,7 +183,7 @@ "MetricExpr": "43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * (1 + = MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UO= PS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L= 3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD= _UOPS_RETIRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "(tma_data_sharing > 0.05) & ((tma_l3_bound > 0= .05) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_contested_accesses, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, @@ -192,7 +192,7 @@ "MetricExpr": "ARITH.FPU_DIV_ACTIVE / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "(tma_divider > 0.2) & ((tma_core_bound > 0.1) = & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%" }, @@ -202,7 +202,7 @@ "MetricExpr": "(1 - MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_= RETIRED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS)) * CYCLE_ACTIVITY.STALL= S_L2_MISS / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "(tma_dram_bound > 0.1) & ((tma_memory_bound > = 0.2) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -211,7 +211,7 @@ "MetricExpr": "(IDQ.ALL_DSB_CYCLES_ANY_UOPS - IDQ.ALL_DSB_CYCLES_4= _UOPS) / tma_info_core_core_clks / 2", "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", - "MetricThreshold": "(tma_dsb > 0.15) & ((tma_fetch_bandwidth > 0.2= ))", + "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, @@ -220,7 +220,7 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "(tma_dsb_switches > 0.05) & ((tma_fetch_latenc= y > 0.1) & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_frontend_dsb_cove= rage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, @@ -229,7 +229,7 @@ "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISS= ES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED) / tm= a_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "(tma_dtlb_load > 0.1) & ((tma_l1_bound > 0.1) = & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_dtlb_store", "ScaleUnit": "100%" }, @@ -238,7 +238,7 @@ "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MI= SSES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED) /= tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "(tma_dtlb_store > 0.05) & ((tma_store_bound > = 0.2) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load", "ScaleUnit": "100%" }, @@ -248,7 +248,7 @@ "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", - "MetricThreshold": "(tma_fb_full > 0.3)", + "MetricThreshold": "tma_fb_full > 0.3", "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", "ScaleUnit": "100%" }, @@ -257,7 +257,7 @@ "MetricExpr": "tma_frontend_bound - tma_fetch_latency", "MetricGroup": "FetchBW;Frontend;TmaL2;TopdownL2;tma_L2_group;tma_= frontend_bound_group;tma_issueFB", "MetricName": "tma_fetch_bandwidth", - "MetricThreshold": "(tma_fetch_bandwidth > 0.2)", + "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" @@ -267,7 +267,7 @@ "MetricExpr": "4 * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE= / tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend= _bound_group", "MetricName": "tma_fetch_latency", - "MetricThreshold": "(tma_fetch_latency > 0.1) & ((tma_frontend_bou= nd > 0.15))", + "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" @@ -277,7 +277,7 @@ "MetricExpr": "tma_x87_use + tma_fp_scalar + tma_fp_vector", "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", - "MetricThreshold": "(tma_fp_arith > 0.2) & ((tma_light_operations = > 0.6))", + "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, @@ -286,7 +286,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "(tma_fp_scalar > 0.1) & ((tma_fp_arith > 0.2) = & ((tma_light_operations > 0.6)))", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -295,7 +295,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "(tma_fp_vector > 0.1) & ((tma_fp_arith > 0.2) = & ((tma_light_operations > 0.6)))", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -304,7 +304,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "(tma_fp_vector_128b > 0.1) & ((tma_fp_vector >= 0.1) & ((tma_fp_arith > 0.2) & ((tma_light_operations > 0.6))))", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -313,7 +313,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "(tma_fp_vector_256b > 0.1) & ((tma_fp_vector >= 0.1) & ((tma_fp_arith > 0.2) & ((tma_light_operations > 0.6))))", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -322,7 +322,7 @@ "MetricExpr": "IDQ_UOPS_NOT_DELIVERED.CORE / tma_info_thread_slots= ", "MetricGroup": "BvFB;BvIO;PGO;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound > 0.15)", + "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" @@ -332,7 +332,7 @@ "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", - "MetricThreshold": "(tma_heavy_operations > 0.1)", + "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%" @@ -342,7 +342,7 @@ "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses > 0.05) & ((tma_fetch_laten= cy > 0.1) & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, @@ -351,14 +351,14 @@ "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "(tma_info_bad_spec_ipmisp_indirect < 1000)" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", "MetricGroup": "Bad;BadSpec;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmispredict", - "MetricThreshold": "(tma_info_bad_spec_ipmispredict < 200)" + "MetricThreshold": "tma_info_bad_spec_ipmispredict < 200" }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", @@ -396,7 +396,7 @@ "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_= UOPS + IDQ.MS_UOPS)", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_frontend_dsb_coverage", - "MetricThreshold": "(tma_info_frontend_dsb_coverage < 0.7) & ((tma= _info_thread_ipc / 4) > 0.35)", + "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & tma_inf= o_thread_ipc / 4 > 0.35", "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_inst_mix_iptb, tma_lcp" }, { @@ -429,7 +429,7 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = FP_ARITH_INST_RETIRED.VECTOR)", "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", - "MetricThreshold": "(tma_info_inst_mix_iparith < 10)", + "MetricThreshold": "tma_info_inst_mix_iparith < 10", "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { @@ -437,7 +437,7 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.128B_PACK= ED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE)", "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", - "MetricThreshold": "(tma_info_inst_mix_iparith_avx128 < 10)", + "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { @@ -445,7 +445,7 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.256B_PACK= ED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE)", "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", - "MetricThreshold": "(tma_info_inst_mix_iparith_avx256 < 10)", + "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { @@ -453,7 +453,7 @@ "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOU= BLE", "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", - "MetricThreshold": "(tma_info_inst_mix_iparith_scalar_dp < 10)", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { @@ -461,7 +461,7 @@ "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SIN= GLE", "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", - "MetricThreshold": "(tma_info_inst_mix_iparith_scalar_sp < 10)", + "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { @@ -469,42 +469,42 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", "MetricGroup": "Branches;Fed;InsType", "MetricName": "tma_info_inst_mix_ipbranch", - "MetricThreshold": "(tma_info_inst_mix_ipbranch < 8)" + "MetricThreshold": "tma_info_inst_mix_ipbranch < 8" }, { "BriefDescription": "Instructions per (near) call (lower number me= ans higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_ipcall", - "MetricThreshold": "(tma_info_inst_mix_ipcall < 200)" + "MetricThreshold": "tma_info_inst_mix_ipcall < 200" }, { "BriefDescription": "Instructions per Floating Point (FP) Operatio= n (lower number means higher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * FP_ARITH_INST_RETIRED.4_= FLOPS + 8 * FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE)", "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_ipflop", - "MetricThreshold": "(tma_info_inst_mix_ipflop < 10)" + "MetricThreshold": "tma_info_inst_mix_ipflop < 10" }, { "BriefDescription": "Instructions per Load (lower number means hig= her occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", "MetricGroup": "InsType", "MetricName": "tma_info_inst_mix_ipload", - "MetricThreshold": "(tma_info_inst_mix_ipload < 3)" + "MetricThreshold": "tma_info_inst_mix_ipload < 3" }, { "BriefDescription": "Instructions per Store (lower number means hi= gher occurrence rate)", "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", "MetricGroup": "InsType", "MetricName": "tma_info_inst_mix_ipstore", - "MetricThreshold": "(tma_info_inst_mix_ipstore < 8)" + "MetricThreshold": "tma_info_inst_mix_ipstore < 8" }, { "BriefDescription": "Instructions per taken branch", "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "(tma_info_inst_mix_iptb < 4 * 2 + 1)", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, = tma_lcp" }, { @@ -629,7 +629,7 @@ "MetricExpr": "(cpu@ITLB_MISSES.WALK_DURATION\\,cmask\\=3D1@ + cpu= @DTLB_LOAD_MISSES.WALK_DURATION\\,cmask\\=3D1@ + cpu@DTLB_STORE_MISSES.WALK= _DURATION\\,cmask\\=3D1@ + 7 * (DTLB_STORE_MISSES.WALK_COMPLETED + DTLB_LOA= D_MISSES.WALK_COMPLETED + ITLB_MISSES.WALK_COMPLETED)) / tma_info_core_core= _clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_page_walks_utilization", - "MetricThreshold": "(tma_info_memory_tlb_page_walks_utilization > = 0.5)" + "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { "BriefDescription": "", @@ -680,7 +680,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "(tma_info_system_ipfarbranch < 1000000)" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", @@ -693,14 +693,14 @@ "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THRE= AD", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_utilization", - "MetricThreshold": "(tma_info_system_kernel_utilization > 0.05)" + "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, { "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD= ", "MetricGroup": "Summary", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))" + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" }, { "BriefDescription": "Total package Power in Watts", @@ -725,7 +725,7 @@ "MetricExpr": "duration_time", "MetricGroup": "Summary", "MetricName": "tma_info_system_time", - "MetricThreshold": "(tma_info_system_time < 1)" + "MetricThreshold": "tma_info_system_time < 1" }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", @@ -769,21 +769,21 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / INST_RETIRED.ANY", "MetricGroup": "Pipeline;Ret;Retire", "MetricName": "tma_info_thread_uoppi", - "MetricThreshold": "(tma_info_thread_uoppi > 1.05)" + "MetricThreshold": "tma_info_thread_uoppi > 1.05" }, { "BriefDescription": "Uops per taken branch", "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "(tma_info_thread_uptb < 4 * 1.5)" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_D= URATION\\,cmask\\=3D1@ + 7 * ITLB_MISSES.WALK_COMPLETED) / tma_info_thread_= clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses > 0.05) & ((tma_fetch_latency= > 0.1) & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, @@ -792,7 +792,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "(tma_l1_bound > 0.1) & ((tma_memory_bound > 0.= 2) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -801,7 +801,7 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.ST= ALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "(tma_l2_bound > 0.05) & ((tma_memory_bound > 0= .2) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -811,7 +811,7 @@ "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_RETIR= ED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS) * CYCLE_ACTIVITY.STALLS_L2_M= ISS / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "(tma_l3_bound > 0.05) & ((tma_memory_bound > 0= .2) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, @@ -821,7 +821,7 @@ "MetricExpr": "29 * (MEM_LOAD_UOPS_RETIRED.L3_HIT * (1 + MEM_LOAD_= UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRE= D.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RET= IRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_UOPS_RET= IRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "(tma_l3_hit_latency > 0.1) & ((tma_l3_bound > = 0.05) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_mem_latency", "ScaleUnit": "100%" }, @@ -830,7 +830,7 @@ "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "(tma_lcp > 0.05) & ((tma_fetch_latency > 0.1) = & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, @@ -839,7 +839,7 @@ "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", - "MetricThreshold": "(tma_light_operations > 0.6)", + "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" @@ -850,7 +850,7 @@ "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_2 + UOPS_DISPATCHED_PORT= .PORT_3 + UOPS_DISPATCHED_PORT.PORT_7 - UOPS_DISPATCHED_PORT.PORT_4) / (2 *= tma_info_core_core_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_load_op_utilization", - "MetricThreshold": "(tma_load_op_utilization > 0.6)", + "MetricThreshold": "tma_load_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Load operations. Sample with: = UOPS_DISPATCHED.PORT_2_3_10", "ScaleUnit": "100%" }, @@ -860,7 +860,7 @@ "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "(tma_lock_latency > 0.2) & ((tma_l1_bound > 0.= 1) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -870,7 +870,7 @@ "MetricExpr": "tma_bad_speculation - tma_branch_mispredicts", "MetricGroup": "BadSpec;BvMS;MachineClears;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueMC;tma_issueSyncxn", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears > 0.1) & ((tma_bad_specula= tion > 0.15))", + "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sh= aring, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_c= ache", "ScaleUnit": "100%" @@ -880,7 +880,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "(tma_mem_bandwidth > 0.2) & ((tma_dram_bound >= 0.1) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", "ScaleUnit": "100%" }, @@ -889,7 +889,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "(tma_mem_latency > 0.1) & ((tma_dram_bound > 0= .1) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", "ScaleUnit": "100%" }, @@ -899,7 +899,7 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_MEM_ANY + RESOURCE_STALLS.SB= ) / (CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC - (UO= PS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if tma_info_thread_ipc > 1.8 else UOPS_EX= ECUTED.CYCLES_GE_2_UOPS_EXEC) - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latenc= y > 0.1 else 0) + RESOURCE_STALLS.SB) * tma_backend_bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", - "MetricThreshold": "(tma_memory_bound > 0.2) & ((tma_backend_bound= > 0.2))", + "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" @@ -909,7 +909,7 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / UOPS_ISSUED.ANY * IDQ.M= S_UOPS / tma_info_thread_slots", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operatio= ns_group;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", - "MetricThreshold": "(tma_microcode_sequencer > 0.05) & ((tma_heavy= _operations > 0.1))", + "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_ope= rations > 0.1", "PublicDescription": "This metric represents fraction of slots the= CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The M= S is used for CISC instructions not supported by the default decoders (like= repeat move strings; or CPUID); or by microcode assists used to address so= me operation modes (like in Floating Point assists). These cases can often = be avoided. Sample with: UOPS_RETIRED.MS. Related metrics: tma_clears_reste= ers, tma_l1_bound, tma_machine_clears, tma_ms_switches", "ScaleUnit": "100%" }, @@ -918,7 +918,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES * tma_branch_resteers = / (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "(tma_mispredicts_resteers > 0.05) & ((tma_bran= ch_resteers > 0.05) & ((tma_fetch_latency > 0.1) & ((tma_frontend_bound > 0= .15))))", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_branch_mispredicts, tma_info_bad_spec_branch_misprediction_cost= ", "ScaleUnit": "100%" }, @@ -927,7 +927,7 @@ "MetricExpr": "(IDQ.ALL_MITE_CYCLES_ANY_UOPS - IDQ.ALL_MITE_CYCLES= _4_UOPS) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", - "MetricThreshold": "(tma_mite > 0.1) & ((tma_fetch_bandwidth > 0.2= ))", + "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck. Sa= mple with: FRONTEND_RETIRED.ANY_DSB_MISS", "ScaleUnit": "100%" }, @@ -936,7 +936,7 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "(tma_ms_switches > 0.05) & ((tma_fetch_latency= > 0.1) & ((tma_frontend_bound > 0.15)))", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mix= ing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -945,7 +945,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_0 / tma_info_core_core_cl= ks", "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", - "MetricThreshold": "(tma_port_0 > 0.6)", + "MetricThreshold": "tma_port_0 > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b,= tma_fp_vector_256b, tma_fp_vector_512b, tma_port_1, tma_port_5, tma_port_6= , tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -954,7 +954,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_1 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", - "MetricThreshold": "(tma_port_1 > 0.6)", + "MetricThreshold": "tma_port_1 > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector= _512b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -963,7 +963,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_2 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_gro= up", "MetricName": "tma_port_2", - "MetricThreshold": "(tma_port_2 > 0.6)", + "MetricThreshold": "tma_port_2 > 0.6", "ScaleUnit": "100%" }, { @@ -971,7 +971,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_3 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_load_op_utilization_gro= up", "MetricName": "tma_port_3", - "MetricThreshold": "(tma_port_3 > 0.6)", + "MetricThreshold": "tma_port_3 > 0.6", "ScaleUnit": "100%" }, { @@ -979,7 +979,7 @@ "MetricExpr": "tma_store_op_utilization", "MetricGroup": "TopdownL6;tma_L6_group;tma_issueSpSt;tma_store_op_= utilization_group", "MetricName": "tma_port_4", - "MetricThreshold": "(tma_port_4 > 0.6)", + "MetricThreshold": "tma_port_4 > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 4 (Store-data). Related metrics: t= ma_split_stores", "ScaleUnit": "100%" }, @@ -988,7 +988,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_5 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", - "MetricThreshold": "(tma_port_5 > 0.6)", + "MetricThreshold": "tma_port_5 > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, t= ma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_6, = tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -997,7 +997,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_6 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", - "MetricThreshold": "(tma_port_6 > 0.6)", + "MetricThreshold": "tma_port_6 > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, = tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5,= tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -1006,7 +1006,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_7 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL6;tma_L6_group;tma_store_op_utilization_gr= oup", "MetricName": "tma_port_7", - "MetricThreshold": "(tma_port_7 > 0.6)", + "MetricThreshold": "tma_port_7 > 0.6", "ScaleUnit": "100%" }, { @@ -1015,7 +1015,7 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES= _GE_1_UOP_EXEC - (UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if tma_info_thread_ip= c > 1.8 else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) - (RS_EVENTS.EMPTY_CYCLES= if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB - RESOURCE_STALLS.= SB - CYCLE_ACTIVITY.STALLS_MEM_ANY) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "(tma_ports_utilization > 0.15) & ((tma_core_bo= und > 0.1) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, @@ -1024,7 +1024,7 @@ "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\,cmask\\=3D1@ / 2 if= #SMT_on else (CYCLE_ACTIVITY.STALLS_TOTAL - (RS_EVENTS.EMPTY_CYCLES if tma= _fetch_latency > 0.1 else 0)) / tma_info_core_core_clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "(tma_ports_utilized_0 > 0.2) & ((tma_ports_uti= lization > 0.15) & ((tma_core_bound > 0.1) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, @@ -1033,7 +1033,7 @@ "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D2@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) / tma_info_core_core_clks= )", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "(tma_ports_utilized_1 > 0.2) & ((tma_ports_uti= lization > 0.15) & ((tma_core_bound > 0.1) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1042,7 +1042,7 @@ "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D3@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clk= s)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "(tma_ports_utilized_2 > 0.15) & ((tma_ports_ut= ilization > 0.15) & ((tma_core_bound > 0.1) & ((tma_backend_bound > 0.2))))= ", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= port_0, tma_port_1, tma_port_5, tma_port_6", "ScaleUnit": "100%" }, @@ -1051,7 +1051,7 @@ "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ / 2 if #SMT_= on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "(tma_ports_utilized_3m > 0.4) & ((tma_ports_ut= ilization > 0.15) & ((tma_core_bound > 0.1) & ((tma_backend_bound > 0.2))))= ", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, @@ -1060,7 +1060,7 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / tma_info_thread_slots", "MetricGroup": "BvUW;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "((tma_retiring > 0.7)|(tma_heavy_operations > = 0.1))", + "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", "MetricgroupNoGroup": "TopdownL1", "PublicDescription": "This category represents fraction of slots u= tilized by useful work i.e. issued uops that eventually get retired. Ideall= y; all pipeline slots would be attributed to the Retiring category. Retiri= ng of 100% would indicate the maximum Pipeline_Width throughput was achieve= d. Maximizing Retiring typically increases the Instructions-per-cycle (see= IPC metric). Note that a high Retiring value does not necessary mean there= is no room for more performance. For example; Heavy-operations or Microco= de Assists are categorized under Retiring. They often indicate suboptimal p= erformance and can often be optimized or avoided. Sample with: UOPS_RETIRED= .SLOTS", "ScaleUnit": "100%" @@ -1071,7 +1071,7 @@ "MetricExpr": "tma_info_memory_load_miss_real_latency * LD_BLOCKS.= NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", - "MetricThreshold": "(tma_split_loads > 0.3)", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, @@ -1080,7 +1080,7 @@ "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_c= ore_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "(tma_split_stores > 0.2) & ((tma_store_bound >= 0.2) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, @@ -1089,7 +1089,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "(tma_sq_full > 0.3) & ((tma_l3_bound > 0.05) &= ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", "ScaleUnit": "100%" }, @@ -1098,7 +1098,7 @@ "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "(tma_store_bound > 0.2) & ((tma_memory_bound >= 0.2) & ((tma_backend_bound > 0.2)))", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, @@ -1107,7 +1107,7 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "(tma_store_fwd_blk > 0.1) & ((tma_l1_bound > 0= .1) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, @@ -1117,7 +1117,7 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "(tma_store_latency > 0.1) & ((tma_store_bound = > 0.2) & ((tma_memory_bound > 0.2) & ((tma_backend_bound > 0.2))))", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, @@ -1126,7 +1126,7 @@ "MetricExpr": "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_cl= ks", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_store_op_utilization", - "MetricThreshold": "(tma_store_op_utilization > 0.6)", + "MetricThreshold": "tma_store_op_utilization > 0.6", "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Store operations. Sample with:= UOPS_DISPATCHED.PORT_7_8", "ScaleUnit": "100%" }, @@ -1135,7 +1135,7 @@ "MetricExpr": "tma_branch_resteers - tma_mispredicts_resteers - tm= a_clears_resteers", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "(tma_unknown_branches > 0.05) & ((tma_branch_r= esteers > 0.05) & ((tma_fetch_latency > 0.1) & ((tma_frontend_bound > 0.15)= )))", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%" }, @@ -1144,7 +1144,7 @@ "MetricExpr": "INST_RETIRED.X87 * tma_info_thread_uoppi / UOPS_RET= IRED.RETIRE_SLOTS", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "(tma_x87_use > 0.1) & ((tma_fp_arith > 0.2) & = ((tma_light_operations > 0.6)))", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" } --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 07DC66FB0 for ; Wed, 26 Mar 2025 00:06:05 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947571; cv=none; b=b3MHrO8EwyOcLpBQ1Kc1Au5EqIHJ9Aeca4RuQ9K6aAzk5kI1bEMBZR65KIHZK8A+0e4yy0D11AD9almUYbWFeyqtMAQyJ/N+I/2i+XG2Ls9+Wp5ISSMcb0Tpm9cJdLySyUZUTIZbCu5MrliMYfGhD+44YjucIkCbuZ9lySQYm6U= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947571; c=relaxed/simple; bh=WsuVbqWh+bJy53QzNA91ImR6ThKQ4xu7bsPgOfOgj+U=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=NagX2eSSSqO6uDljiqLLeL3gtL9ig7N7Rt8KwRWz6cEYTfRL3s8RlVVHO2kTXobKfZrRDgm85gFhdlSMj5dgnkb0KlkZcQp7NpKDcUFTRMyk7eDcfhyWHkHh0Xj/IHGJ7DfD+oTB9c5gFU75TaNBm0q5IqaJ2Zt4bHBZggjM778= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=ZT+og4F5; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="ZT+og4F5" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2240a96112fso169711635ad.2 for ; Tue, 25 Mar 2025 17:06:05 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947565; x=1743552365; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=J5Ova8GyhEIII8UmDuNM8qm3xriqxYZGD7PNX7po890=; b=ZT+og4F5kptlV+AcXuBWx0nX4TfMUfgoKGIJr35+dt3RSW+sTMGSkyYgd5mlKrLgLo cZVUIE9N8feCk3Spjemhuy6rICzb71uJZ2m0tWGohi7nF5k5nHk7yUTdTOxksf/LZu6l UuRRXLkHGNsKo95MxhqcCPsbKvkm3wwuVvCqvWnAIUCMgycHdUocDT27TzPb4btGP2ph SHhnRBoSvk4axQAVGyedr9nNyG3HV90IdpLWvmLmnOzS1812DBgAUlMxKaIzv3/nlSEX ubwiiBhkWgOWgjhGNyyhtKNoHftYgGC/lrKgIdW+rkGiriprGKzQajGksTX/hiOsuope paGA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947565; x=1743552365; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=J5Ova8GyhEIII8UmDuNM8qm3xriqxYZGD7PNX7po890=; b=pl+7YROfOVF8yRrH7AEe2nKKIp1uxynVYVbvUM0tCfDcepscViw2VOGo+5HuHqIuPB V/L96jsySumAeZZA7yk8CJcLscIT4QdNY6fHEpfEF9N73zSnVJLhidM7pxTqDxCmAf9u qcS8QhncwMLyC0wLW8EU2ycTrryED/i+WSkcwDKoJOsQLKk84Xmp1XkwvZITxTbw8nWQ aw1RSDc/qdGnrCCaBB4RLzBifF5JOWJYckSGH4xq5NHUdWXX7/GKfK6oHgGdYTnyh3o8 uU4qvDPTjwzJKBRDMFC1SprwiSI81NcGBuNGxu3UNeePvA66wP9gOU34pUX7I1qsuhQ6 YIxA== X-Forwarded-Encrypted: i=1; AJvYcCUNz2WqUVRg946E1JyWFUu8A0mnHF33c9I0MztKfVUEbAX6669oB7oIiaH45W9YpX9lAfSP9M69BOOUaEg=@vger.kernel.org X-Gm-Message-State: AOJu0YyM0p2c+dAelfwwWyMzD1hn9cdMUai42kwj1v8jfsgpxr4+TUVS /TYBdEuBk4XYHPj+qXiCDZoL+lBnyTgbioMxjJfHoJOHPaFTlIRrt8CsBhSsCocr76i4ZjGPzgK 3a0iS3A== X-Google-Smtp-Source: AGHT+IHSX0MFW9CNsHAB2T93xlrh50SgH/8+jnvA2hbiYftrTneKPLYpDNsBcq2FnBY9GBQErp9wech3Lb/N X-Received: from pfwy21.prod.google.com ([2002:a05:6a00:1c95:b0:730:7c5b:2e2b]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:1409:b0:737:e73:f64b with SMTP id d2e1a72fcca58-7390593b7ffmr27324906b3a.1.1742947565225; Tue, 25 Mar 2025 17:06:05 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:05 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-8-irogers@google.com> Subject: [PATCH v3 07/35] perf vendor events: Update broadwellx metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/broadwellx/bdx-metrics.json | 268 +++++++++--------- 1 file changed, 133 insertions(+), 135 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/broadwellx/bdx-metrics.json b/t= ools/perf/pmu-events/arch/x86/broadwellx/bdx-metrics.json index 8016202bad1f..5d06a3f72be2 100644 --- a/tools/perf/pmu-events/arch/x86/broadwellx/bdx-metrics.json +++ b/tools/perf/pmu-events/arch/x86/broadwellx/bdx-metrics.json @@ -276,12 +276,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", @@ -294,8 +294,8 @@ "MetricExpr": "66 * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_= slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y_WB_ASSIST", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, { @@ -306,7 +306,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -316,7 +316,7 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { @@ -327,7 +327,7 @@ "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_mispredicts_resteers", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers", "ScaleUnit": "100%" }, { @@ -335,8 +335,8 @@ "MetricExpr": "12 * (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS= .COUNT + BACLEARS.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -345,8 +345,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -354,7 +354,7 @@ "MetricExpr": "MACHINE_CLEARS.COUNT * tma_branch_resteers / (BR_MI= SP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Rel= ated metrics: tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches", "ScaleUnit": "100%" }, @@ -364,8 +364,8 @@ "MetricExpr": "(60 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * (1 = + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_= UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS= _L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LO= AD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_D= RAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RET= IRED.REMOTE_FWD))) + 43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * (1 + ME= M_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS= _RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_= HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_U= OPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM = + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED= .REMOTE_FWD)))) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MIS= S. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears= , tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { @@ -376,7 +376,7 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { @@ -385,8 +385,8 @@ "MetricExpr": "43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * (1 + = MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UO= PS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L= 3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD= _UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRA= M + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIR= ED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, t= ma_false_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_contested_accesses, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -394,8 +394,8 @@ "MetricExpr": "ARITH.FPU_DIV_ACTIVE / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.FPU_DIV_ACTIVE", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -404,8 +404,8 @@ "MetricExpr": "(1 - MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_= RETIRED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS)) * CYCLE_ACTIVITY.STALL= S_L2_MISS / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS_PS", "ScaleUnit": "100%" }, { @@ -414,7 +414,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -422,26 +422,26 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Related metrics: tma_fetch_bandw= idth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISS= ES.WALK_DURATION\\,cmask\\=3D0x1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED) / = tma_info_thread_clks", + "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + cpu@DTLB_LOAD_MISS= ES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_LOAD_MISSES.WALK_COMPLETED) / tm= a_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MI= SSES.WALK_DURATION\\,cmask\\=3D0x1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED)= / tma_info_thread_clks", + "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + cpu@DTLB_STORE_MI= SSES.WALK_DURATION\\,cmask\\=3D1@ + 7 * DTLB_STORE_MISSES.WALK_COMPLETED) /= tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -449,18 +449,18 @@ "MetricExpr": "(200 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_= HITM + 60 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE) / tma_info= _thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3= _HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_= RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LL= C_MISS.REMOTE_HITM. Related metrics: tma_contested_accesses, tma_data_shari= ng, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -489,7 +489,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -497,8 +497,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports= _utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -506,8 +506,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -515,8 +515,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -524,8 +524,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -535,33 +535,33 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound.", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses.", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -572,7 +572,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -593,11 +593,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -622,7 +622,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -640,7 +640,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -648,7 +648,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -656,7 +656,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -664,7 +664,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -672,7 +672,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -714,7 +714,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, = tma_lcp" }, { @@ -842,14 +842,14 @@ "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D0x1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -890,14 +890,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -908,14 +907,14 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\= =3D0x182@ / cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=3D0x182@", + "MetricExpr": "UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\=3D0x18= 2@ / UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\=3D0x182\\,thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" }, { "BriefDescription": "Average latency of data read request to exter= nal memory (in nanoseconds)", - "MetricExpr": "1e9 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filte= r_opc\\=3D0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=3D0x18= 2@) / (tma_info_system_socket_clks / tma_info_system_time)", + "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / tma_info_system_time)", "MetricGroup": "Mem;MemoryLat;SoC", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetche= s. ([RKL+]memory-controller only)" @@ -965,7 +964,7 @@ "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -974,15 +973,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1008,14 +1006,14 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", - "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_D= URATION\\,cmask\\=3D0x1@ + 7 * ITLB_MISSES.WALK_COMPLETED) / tma_info_threa= d_clks", + "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + cpu@ITLB_MISSES.WALK_D= URATION\\,cmask\\=3D1@ + 7 * ITLB_MISSES.WALK_COMPLETED) / tma_info_thread_= clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_M= ISSES.WALK_COMPLETED", "ScaleUnit": "100%" }, @@ -1024,8 +1022,8 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metri= cs: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_m= s_switches, tma_ports_utilized_1", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -1033,8 +1031,8 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.ST= ALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT_PS", "ScaleUnit": "100%" }, { @@ -1043,8 +1041,8 @@ "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_RETIR= ED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS) * CYCLE_ACTIVITY.STALLS_L2_M= ISS / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { @@ -1053,8 +1051,8 @@ "MetricExpr": "41 * (MEM_LOAD_UOPS_RETIRED.L3_HIT * (1 + MEM_LOAD_= UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRE= D.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RET= IRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_UOPS_L3_= MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_L= OAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE= _FWD))) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: = tma_branch_resteers, tma_mem_latency, tma_store_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS. Related metric= s: tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1062,18 +1060,18 @@ "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage,= tma_info_inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1091,8 +1089,8 @@ "MetricExpr": "200 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM * (= 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOA= D_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UO= PS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_= LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE= _DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_R= ETIRED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.LOCAL_DRAM", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.LOCAL_DRAM_PS", "ScaleUnit": "100%" }, { @@ -1101,8 +1099,8 @@ "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, { @@ -1118,10 +1116,10 @@ }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", "ScaleUnit": "100%" }, @@ -1130,7 +1128,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", "ScaleUnit": "100%" }, @@ -1142,7 +1140,7 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { @@ -1159,8 +1157,8 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES * tma_branch_resteers = / (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS.COUNT + BACLEARS.ANY)", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Related metrics: tma_branch_mispredicts", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Related metrics: tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, { @@ -1169,7 +1167,7 @@ "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", "ScaleUnit": "100%" }, { @@ -1177,8 +1175,8 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mix= ing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, { @@ -1187,7 +1185,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, = tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1196,7 +1194,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_= utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1232,7 +1230,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_sc= alar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tm= a_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1241,7 +1239,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, t= ma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1259,43 +1257,43 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_TOTAL + UOPS_EXECUTED.CYCLES= _GE_1_UOP_EXEC - (UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC if tma_info_thread_ip= c > 1.8 else UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) - (RS_EVENTS.EMPTY_CYCLES= if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB - RESOURCE_STALLS.= SB - CYCLE_ACTIVITY.STALLS_MEM_ANY) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed no uops on any execution port (Logical Processor cycles since ICL= , Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\=3D0x1\\,cmask\\=3D0= x1@ / 2 if #SMT_on else CYCLE_ACTIVITY.STALLS_TOTAL - (RS_EVENTS.EMPTY_CYCL= ES if tma_fetch_latency > 0.1 else 0)) / tma_info_core_core_clks", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\,cmask\\=3D1@ / 2 if= #SMT_on else (CYCLE_ACTIVITY.STALLS_TOTAL - (RS_EVENTS.EMPTY_CYCLES if tma= _fetch_latency > 0.1 else 0)) / tma_info_core_core_clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re the CPU executed total of 1 uop per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x2@) / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_= GE_1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) / tma_info_core_core_c= lks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D2@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 1_UOP_EXEC - UOPS_EXECUTED.CYCLES_GE_2_UOPS_EXEC) / tma_info_core_core_clks= )", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 2 uops per cycle on all execution ports (Logical Process= or cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x3@) / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_= GE_2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_= clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D3@) / 2 if #SMT_on else (UOPS_EXECUTED.CYCLES_GE_= 2_UOPS_EXEC - UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clk= s)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ / 2 if #SM= T_on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clks", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ / 2 if #SMT_= on else UOPS_EXECUTED.CYCLES_GE_3_UOPS_EXEC) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1304,8 +1302,8 @@ "MetricExpr": "(200 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM *= (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_L= OAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + ME= M_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMO= TE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS= _RETIRED.REMOTE_FWD))) + 180 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD * = (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LO= AD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM= _LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOT= E_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_= RETIRED.REMOTE_FWD)))) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM= , MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_contested_= accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_UOPS_L3_M= ISS_RETIRED.REMOTE_HITM_PS;MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD_PS. Rel= ated metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, = tma_machine_clears", "ScaleUnit": "100%" }, { @@ -1313,8 +1311,8 @@ "MetricExpr": "310 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM * = (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LO= AD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM= _LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOT= E_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_= RETIRED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { @@ -1334,7 +1332,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1342,8 +1340,8 @@ "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_c= ore_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1351,7 +1349,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", "ScaleUnit": "100%" }, @@ -1360,8 +1358,8 @@ "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1369,8 +1367,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1379,8 +1377,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1396,7 +1394,7 @@ "MetricExpr": "tma_branch_resteers - tma_mispredicts_resteers - tm= a_clears_resteers", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1405,8 +1403,8 @@ "MetricExpr": "INST_RETIRED.X87 * tma_info_thread_uoppi / UOPS_RET= IRED.RETIRE_SLOTS", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id AF1DCD27E for ; Wed, 26 Mar 2025 00:06:14 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947585; cv=none; b=HBIemknL0IERBiiqjiLwEmbIORjwcFGbFboBF/7QXk3Bs4NQFJ3Ej3HQy8jhprixkovUmxHzDB5o6HhCU+Wo08npze4ulWt5WPZp7nXdaTznNCg0Mek1+7gDDIvOjRPPd6sCwQr2h55C+0Ct1UKHteE4TZjLU8paeO5oJD+DPrs= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947585; c=relaxed/simple; bh=EGxdMtxUGHyCcu+Hd0Y6QDl5Ws+H7DHDAF99GmO0dnc=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=fTWn4nQ5jRMapK7zwkfMCrRTtgheMloZsPl9k95Sfq3Fc+Yjt7PP7nxY6LYFKnCZrIPjBGGOEVSgK68oMzjukuucrO8eFNkPZ90velIDio5W8UfCvUV46mCx3RBWsMS8HSJU92M6IGvMk+Fe1gwdMmuIWBxFvA4XC/RcamBz8kc= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=OdfBFizb; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="OdfBFizb" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff6af1e264so17528558a91.3 for ; Tue, 25 Mar 2025 17:06:14 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947574; x=1743552374; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=n6Q8Z4WlSxJWH025xs6NtydM/dPbwQWyqyzyunuITcw=; b=OdfBFizbfTenInu3my8t71TatobM4Wr1oBG2XLQIwoA6pfV22yVrEpnQAunZVx8DRE lswvozC6uGETKQUREjz3K0aJAKN9cweZB2Q55gbqw6YzQmcApwl34ZzyIs34T4QSAxhH qJMCWS1IfRWiVSjvWcrYvf22+sFIGpBd/FAxMOv5FV4P5WSdVE/RVCNaxohoSXHqHnH+ Cao0vyI6uS/PmJop3rsVQGcd0AuEDhS0FISaZoDSXUZ2dkUPzUcUM4dk8DWY/SmtWOA/ FUtDzG8aNjgu3zmevTS8ssL1bdUZeHrFv8FT4mletus+MXV2s7YBz1jECUQbPmmCBzA9 qbIw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947574; x=1743552374; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=n6Q8Z4WlSxJWH025xs6NtydM/dPbwQWyqyzyunuITcw=; b=CjTiQV19Je7+GRrje0dlymHe/W2KIdIX/BRg8dy9tCjP+suzogOTU3dKMD7PcEJbCh 9byHIwTfNydathxbsgcsJ8J9efNyPTagJDxDScR+SQz6EfzC6FHB4sK9VqI/pDfm+VM0 Ovx1ajih9JYQh5osTeTnfZ6mqkCVm/SmZTEpCFJcGltRf3w+/lVNouNwciDk8oRJPW2E y7qw9T8Uj2e1Hu29Pei9jUsPsKLlvlX2Oun61fSP+i4rcifkzlTItaKrSodoBw7IpTHV Yv4XBZINRgPH2WMNQosReeVCU/zz7JimEGG3TGLh0ntK4hI+UAMYbvkTes8Yxo0t/+HW IuiA== X-Forwarded-Encrypted: i=1; AJvYcCVr3w81bVaN+AKDFoNHxlHnx8jVrd0MMzXcc5EFIL17/U4ZsU50Do6pmcEcmR/grqMTVB8R/e42BkiEQAA=@vger.kernel.org X-Gm-Message-State: AOJu0Ywd9jL+dx6kMRBeO0ufWgMdfUHkTQfX/iBx19EWOPmLsSGjsYwT 6s5HIKwKxRVkvmqJ6qgodDWIOXUn4bFPWfLjrM5HR5IjP/qtxQUBdpMqZgkV+LHT06dntM75Sxc d8gtZ5g== X-Google-Smtp-Source: AGHT+IFRyPEh7YsnmIMF21VyuEWhHIiIR8VV3JaHHcWqPbrsgs3xnnEUlf9B8eqtBcjLdls+uJ0kKpAFUPvX X-Received: from pjbsj16.prod.google.com ([2002:a17:90b:2d90:b0:301:a339:b558]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:1650:b0:2ee:fa0c:cebc with SMTP id 98e67ed59e1d1-3030fea0d2amr27811858a91.20.1742947574002; Tue, 25 Mar 2025 17:06:14 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:06 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-9-irogers@google.com> Subject: [PATCH v3 08/35] perf vendor events: Update cascadelakex events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/cascadelakex/cache.json | 404 ++++++++++++++++++ .../arch/x86/cascadelakex/clx-metrics.json | 389 +++++++++-------- .../arch/x86/cascadelakex/other.json | 404 ------------------ 3 files changed, 598 insertions(+), 599 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/cascadelakex/cache.json b/tools= /perf/pmu-events/arch/x86/cascadelakex/cache.json index 8bad700ff8ea..d113c14aa7c9 100644 --- a/tools/perf/pmu-events/arch/x86/cascadelakex/cache.json +++ b/tools/perf/pmu-events/arch/x86/cascadelakex/cache.json @@ -1,4 +1,78 @@ [ + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", + "SampleAfterValue": "2000003", + "UMask": "0x20" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDM", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDM", + "SampleAfterValue": "2000003", + "UMask": "0x10" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", + "SampleAfterValue": "2000003", + "UMask": "0x2" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITI", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITI", + "SampleAfterValue": "2000003", + "UMask": "0x1" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", + "SampleAfterValue": "2000003", + "UMask": "0x40" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDM", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDM", + "SampleAfterValue": "2000003", + "UMask": "0x8" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", + "SampleAfterValue": "2000003", + "UMask": "0x4" + }, + { + "BriefDescription": "Counts number of cache lines that are dropped= and not written back to L3 as they are deemed to be less likely to be reus= ed shortly", + "Counter": "0,1,2,3", + "EventCode": "0xFE", + "EventName": "IDI_MISC.WB_DOWNGRADE", + "PublicDescription": "Counts number of cache lines that are droppe= d and not written back to L3 as they are deemed to be less likely to be reu= sed shortly.", + "SampleAfterValue": "100003", + "UMask": "0x4" + }, + { + "BriefDescription": "Counts number of cache lines that are allocat= ed and written back to L3 with the intention that they are more likely to b= e reused shortly", + "Counter": "0,1,2,3", + "EventCode": "0xFE", + "EventName": "IDI_MISC.WB_UPGRADE", + "PublicDescription": "Counts number of cache lines that are alloca= ted and written back to L3 with the intention that they are more likely to = be reused shortly.", + "SampleAfterValue": "100003", + "UMask": "0x2" + }, { "BriefDescription": "L1D data line replacements", "Counter": "0,1,2,3", @@ -2343,6 +2417,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all demand code reads have any respons= e type.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.L3_HIT.ANY_SNOOP OCR.DEMAND_CODE_RD.L3_HIT.ANY_SNOOP", "Counter": "0,1,2,3", @@ -2703,6 +2787,116 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80400004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80400004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDE= D", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HITM_OTHER_CORE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HITM_OTHER_CORE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1000020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x800020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_F= WD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x400020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.SNOOP_MISS", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x200020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand code reads", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80020004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads have any response ty= pe.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.L= 3_HIT.ANY_SNOOP OCR.DEMAND_DATA_RD.L3_HIT.ANY_SNOOP", "Counter": "0,1,2,3", @@ -3063,6 +3257,116 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80400001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.SNOOP_NONE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80400001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDE= D", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HITM_OTHER_CORE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HITM_OTHER_CORE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1000020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x800020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_F= WD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x400020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.NO_SNOOP_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.SNOOP_MISS", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x200020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80020001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) have any= response type.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.L3_HIT.ANY_SNOOP OCR.DEMAND_RFO.L3_HIT.ANY_SNOOP", "Counter": "0,1,2,3", @@ -3423,6 +3727,106 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80400002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NONE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80400002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.ANY_SNOOP", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.ANY_SNOOP", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F80020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HITM_OTHER_CORE", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HITM_OTHER_CORE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1000020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x800020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x400020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.NO_SNOOP_NEEDED", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.NO_SNOOP_NEEDED", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs)", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.SNOOP_MISS", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x200020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all demand data writes (RFOs)", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.SNOOP_NONE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x80020002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts any other requests OCR.OTHER.L3_HIT.AN= Y_SNOOP OCR.OTHER.L3_HIT.ANY_SNOOP", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/cascadelakex/clx-metrics.json b= /tools/perf/pmu-events/arch/x86/cascadelakex/clx-metrics.json index 5729b93a9c68..6485b565acbc 100644 --- a/tools/perf/pmu-events/arch/x86/cascadelakex/clx-metrics.json +++ b/tools/perf/pmu-events/arch/x86/cascadelakex/clx-metrics.json @@ -313,12 +313,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -330,7 +330,7 @@ "MetricExpr": "34 * (FP_ASSIST.ANY + OTHER_ASSISTS.ANY) / tma_info= _thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, @@ -341,7 +341,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -351,12 +351,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -371,7 +371,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -379,7 +379,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_dtlb_store)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound= + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_= store_latency / (tma_store_latency + tma_false_sharing + tma_split_stores += tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency)) + tma_memory_bound * (tma_store_bound / (tma_dram_bou= nd + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_= store_latency / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tm= a_store_latency)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -387,22 +387,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_b= ranches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tm= a_ms_switches + tma_lcp + tma_dsb_switches)) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_b= ranches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + t= ma_itlb_misses + tma_lcp + tma_ms_switches)) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_mispredicts_resteers = + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_i= tlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_clears_resteers + tma= _mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma= _dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -410,7 +410,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -418,7 +418,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_dtlb_store - tma_store_latency)) + tma_machine_clears * (1 = - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency - tma_store_latency)) + tma_machine_clears * (1 = - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -426,7 +426,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -438,10 +438,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -463,8 +463,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -472,8 +472,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -481,7 +481,7 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, @@ -490,7 +490,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -498,33 +498,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((47.5 * tma_info_system_core_frequency - 3.5 * tma= _info_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (OCR.DE= MAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / (OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER= _CORE + OCR.DEMAND_DATA_RD.L3_HIT.HIT_OTHER_CORE_FWD))) + (47.5 * tma_info_= system_core_frequency - 3.5 * tma_info_system_core_frequency) * MEM_LOAD_L3= _HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L= 1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(44 * tma_info_system_core_frequency * (MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM * (OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / (OCR.= DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OCR.DEMAND_DATA_RD.L3_HIT.HIT_OTHER= _CORE_FWD))) + 44 * tma_info_system_core_frequency * MEM_LOAD_L3_HIT_RETIRE= D.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)= / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -535,25 +535,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(47.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= L3_HIT_RETIRED.XSNP_HITM * (1 - OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE /= (OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OCR.DEMAND_DATA_RD.L3_HIT.HIT= _OTHER_CORE_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MIS= S / 2) / tma_info_thread_clks", + "MetricExpr": "44 * tma_info_system_core_frequency * (MEM_LOAD_L3_= HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (1 - OCR.DEMAND_= DATA_RD.L3_HIT.HITM_OTHER_CORE / (OCR.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE= + OCR.DEMAND_DATA_RD.L3_HIT.HIT_OTHER_CORE_FWD))) * (1 + MEM_LOAD_RETIRED.= FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma= _remote_cache", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -562,7 +562,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -572,7 +572,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -582,7 +582,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -590,27 +590,27 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -619,18 +619,18 @@ "MetricExpr": "(110 * tma_info_system_core_frequency * (OCR.DEMAND= _RFO.L3_MISS.REMOTE_HITM + OCR.PF_L2_RFO.L3_MISS.REMOTE_HITM) + 47.5 * tma_= info_system_core_frequency * (OCR.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE + OCR.P= F_L2_RFO.L3_HIT.HITM_OTHER_CORE)) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM, OCR.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE. Related metrics: = tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data_sha= ring, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tm= a_data_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -640,7 +640,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -650,7 +650,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -670,7 +670,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -679,7 +679,7 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { @@ -687,17 +687,17 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { "BriefDescription": "This metric approximates arithmetic floating-= point (FP) vector uops fraction the CPU has retired aggregated across all v= ector widths", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umas= k\\=3D0xFC@ / UOPS_RETIRED.RETIRE_SLOTS", + "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umas= k\\=3D0xfc@ / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -706,7 +706,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -715,7 +715,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -724,7 +724,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -735,35 +735,35 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / U= OPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "(UOPS_RETIRED.RETIRE_SLOTS + UOPS_RETIRED.MACRO_FUS= ED - INST_RETIRED.ANY) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", - "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D0x1\\,edge\\=3D0x1@) / tma_info_thread_clks", + "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D1\\,edge@) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -774,11 +774,11 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -803,7 +803,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= )))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= )))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -812,7 +812,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -820,10 +820,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -852,7 +853,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -877,14 +878,14 @@ }, { "BriefDescription": "Actual per-core usage of the Floating Point n= on-X87 execution units (regardless of precision or vector-width)", - "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + cpu@FP_ARITH_INST_R= ETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xFC@) / (2 * tma_info_core_core_clks= )", + "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + cpu@FP_ARITH_INST_R= ETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xfc@) / (2 * tma_info_core_core_clks= )", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -897,20 +898,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHE= S.COUNT", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@ + 2", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -946,7 +947,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -961,11 +962,11 @@ { "BriefDescription": "Instructions per FP Arithmetic instruction (l= ower number means higher occurrence rate)", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xFC@)", + "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xfc@)", "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -973,7 +974,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -981,7 +982,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -989,7 +990,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -997,7 +998,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -1005,7 +1006,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1061,7 +1062,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1248,8 +1249,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1270,12 +1271,12 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ASSIST.ANY + OTHER_ASSISTS.A= NY)", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1331,14 +1332,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1356,7 +1356,7 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1386,7 +1386,7 @@ "MetricExpr": "(CORE_POWER.LVL0_TURBO_LICENSE / 2 / tma_info_core_= core_clks if #SMT_on else CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_cor= e_clks)", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1394,7 +1394,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1402,7 +1402,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1436,7 +1436,7 @@ "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1445,15 +1445,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1479,15 +1478,15 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1495,7 +1494,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1504,17 +1503,17 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates how often the CPU was s= talled due to L2 cache accesses by loads", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D0x1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2= _MISS) / tma_info_thread_clks)", + "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_M= ISS) / tma_info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1523,7 +1522,7 @@ "MetricExpr": "3.5 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1532,17 +1531,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(20.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETI= RED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "17 * tma_info_system_core_frequency * (MEM_LOAD_RET= IRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2))= / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1550,18 +1549,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1579,7 +1578,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1587,39 +1586,39 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from local memory", - "MetricExpr": "(80 * tma_info_system_core_frequency - 20.5 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "59.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _MISS_RETIRED.LOCAL_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1628,7 +1627,7 @@ "MetricExpr": "(12 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (11= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1645,10 +1644,10 @@ }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1657,7 +1656,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1669,11 +1668,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1695,7 +1694,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1709,12 +1708,12 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { @@ -1722,7 +1721,7 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1732,7 +1731,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1740,8 +1739,8 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / UOPS_RETI= RED.RETIRE_SLOTS", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP_PS", "ScaleUnit": "100%" }, { @@ -1754,19 +1753,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1775,7 +1774,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vecto= r_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1784,7 +1783,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tm= a_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1820,7 +1819,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_sc= alar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_= 512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1829,7 +1828,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector= _512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1846,8 +1845,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1855,8 +1854,8 @@ "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_cl= ks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1864,7 +1863,7 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CO= RE_CYCLES_GE_2) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1873,35 +1872,35 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CO= RE_CYCLES_GE_3) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", "MetricExpr": "(UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else= UOPS_EXECUTED.CORE_CYCLES_GE_3) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote cache in other socket= s including synchronizations issues", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "((110 * tma_info_system_core_frequency - 20.5 * tma= _info_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + (110 = * tma_info_system_core_frequency - 20.5 * tma_info_system_core_frequency) *= MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_= LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(89.5 * tma_info_system_core_frequency * MEM_LOAD_L= 3_MISS_RETIRED.REMOTE_HITM + 89.5 * tma_info_system_core_frequency * MEM_LO= AD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RE= TIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote memory", - "MetricExpr": "(147.5 * tma_info_system_core_frequency - 20.5 * tm= a_info_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * (1 += MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_= clks", + "MetricExpr": "127 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { @@ -1919,7 +1918,7 @@ "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clk= s", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related me= trics: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1928,8 +1927,8 @@ "MetricExpr": "40 * ROB_MISC_EVENTS.PAUSE_INST / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: ROB_MISC_EVENTS.P= AUSE_INST", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUS= E_INST", "ScaleUnit": "100%" }, { @@ -1939,7 +1938,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1947,8 +1946,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1956,7 +1955,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1965,8 +1964,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1974,8 +1973,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1984,8 +1983,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 11 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -2001,7 +2000,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -2009,31 +2008,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2041,7 +2040,7 @@ "MetricExpr": "9 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -2050,8 +2049,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/cascadelakex/other.json b/tools= /perf/pmu-events/arch/x86/cascadelakex/other.json index f25693b17b8b..51833bce994e 100644 --- a/tools/perf/pmu-events/arch/x86/cascadelakex/other.json +++ b/tools/perf/pmu-events/arch/x86/cascadelakex/other.json @@ -35,62 +35,6 @@ "SampleAfterValue": "200003", "UMask": "0x40" }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", - "SampleAfterValue": "2000003", - "UMask": "0x20" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDM", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDM", - "SampleAfterValue": "2000003", - "UMask": "0x10" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", - "SampleAfterValue": "2000003", - "UMask": "0x2" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITI", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITI", - "SampleAfterValue": "2000003", - "UMask": "0x1" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", - "SampleAfterValue": "2000003", - "UMask": "0x40" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDM", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDM", - "SampleAfterValue": "2000003", - "UMask": "0x8" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", - "SampleAfterValue": "2000003", - "UMask": "0x4" - }, { "BriefDescription": "Number of hardware interrupts received by the= processor.", "Counter": "0,1,2,3", @@ -100,24 +44,6 @@ "SampleAfterValue": "203", "UMask": "0x1" }, - { - "BriefDescription": "Counts number of cache lines that are dropped= and not written back to L3 as they are deemed to be less likely to be reus= ed shortly", - "Counter": "0,1,2,3", - "EventCode": "0xFE", - "EventName": "IDI_MISC.WB_DOWNGRADE", - "PublicDescription": "Counts number of cache lines that are droppe= d and not written back to L3 as they are deemed to be less likely to be reu= sed shortly.", - "SampleAfterValue": "100003", - "UMask": "0x4" - }, - { - "BriefDescription": "Counts number of cache lines that are allocat= ed and written back to L3 with the intention that they are more likely to b= e reused shortly", - "Counter": "0,1,2,3", - "EventCode": "0xFE", - "EventName": "IDI_MISC.WB_UPGRADE", - "PublicDescription": "Counts number of cache lines that are alloca= ted and written back to L3 with the intention that they are more likely to = be reused shortly.", - "SampleAfterValue": "100003", - "UMask": "0x2" - }, { "BriefDescription": "OCR.ALL_DATA_RD.ANY_RESPONSE have any respons= e type.", "Counter": "0,1,2,3", @@ -668,336 +594,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all demand code reads have any respons= e type.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80400004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80400004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE_= RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDE= D", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HITM_OTHER_CORE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HITM_OTHER_CORE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1000020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x800020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_F= WD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x400020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads OCR.DEMAND_CODE= _RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.SNOOP_MISS", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x200020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand code reads", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SUPPLIER_NONE.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80020004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads have any response ty= pe.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80400001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.SNOOP_NONE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80400001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.P= MM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDE= D", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HITM_OTHER_CORE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HITM_OTHER_CORE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1000020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x800020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.HIT_OTHER_CORE_NO_F= WD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x400020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads OCR.DEMAND_DATA_RD.= SUPPLIER_NONE.NO_SNOOP_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.NO_SNOOP_NEEDED", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.SNOOP_MISS", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x200020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SUPPLIER_NONE.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80020001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) have any= response type.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80400002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NONE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80400002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEMA= ND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.PMM_HIT_LOCAL_PMM.SNOOP_NOT_NEEDED", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.ANY_SNOOP", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.ANY_SNOOP", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F80020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HITM_OTHER_CORE", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HITM_OTHER_CORE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1000020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_FWD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x800020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.HIT_OTHER_CORE_NO_FWD", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x400020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs) OCR.DEM= AND_RFO.SUPPLIER_NONE.NO_SNOOP_NEEDED", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.NO_SNOOP_NEEDED", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs)", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.SNOOP_MISS", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x200020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all demand data writes (RFOs)", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SUPPLIER_NONE.SNOOP_NONE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x80020002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts any other requests have any response t= ype.", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 86CA1286A9 for ; Wed, 26 Mar 2025 00:06:23 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947585; cv=none; b=F5iS2sokTpVRu2hFAdsHOhVnBgve1vOgiQZbmqYmMlNzUUJCjkkKFJ1rsq6RNSMYj3aywfWRQgERtpMLQrbVBNZxNhPGIYTX2bZMUA/hncxV1ufe6UqBh37XIVcpG3FndD/WUSNDgtzkU4ZtMBn4D19iYPj3UJvuUuU+XCtajL4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947585; c=relaxed/simple; bh=jj7cgVY1dHCdbl7CToIOpx+DtAOgzXi+ZwBseYtASCc=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=kgCsOTe+j3zox4HdtKUaHXVbbgPtDdLr4VP5Bk4/plsis54vA6x81hFA7FS0JZqU/Ugd+1ONJBprZQoSqPQOVO2Oiixdp3OqcPlPUrOL5aARXP29PCmWCEZEcqTyglIugZGIo18pKO1tACBiJSS86ZJ9TmoEBy9XwqtilJFe04c= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=Ubqg7UTq; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="Ubqg7UTq" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff7f9a0b9bso10770173a91.0 for ; Tue, 25 Mar 2025 17:06:23 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947583; x=1743552383; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=6tXKzMdne0PcZ1yEwuy/OWzqt9sFMlDxjtYmEokV0uc=; b=Ubqg7UTq9r/Hdkk87250xpWbi7HOpyje7lAqkhYCWr3aUfGxIKht2Zio1DI+NNfWoO A49DBDRO/UVeJyxzoltRyhnRahUC7zava/5v/30JMUWMLmgDQgAcCz+cImgURcha1QY5 1BDpGaMKFR9U3SJ3Q4bErQffH8GTvDhstr8D2FwAidT3aJUAHGk542EVu/YJhEIccfVN jiivm4s2VkEV0f6cf/ZOWFRD8+6821NAv0mOKNdXkOLIwmbHDD3nx4Bi61uXo2k5TP01 uNsHwHoIXpcr+cujXcvTl5GfXSTgVQ19JFzm2n95iaCDygm9coRYz1PWrfsP6JD+JUrB 2yQg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947583; x=1743552383; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=6tXKzMdne0PcZ1yEwuy/OWzqt9sFMlDxjtYmEokV0uc=; b=UQK2gnS6sSXK3KkTDqz7OnwFvDoBkuSbBrBASpmhN8GHgiMf8Bi3GPw2E4FyLz3qwG lzlORRU1cateywIFR+EhCNNbPnKM5JzkqK/rdEJgwQvnIZbvhZKKhkHkpnoaESoVqvcu sX+jkZ69KQONMVxOCVqeVRtk7gOwC88SCiupyHWAmXUQb1aRKXgqdFiopA7sDSeKVJik oGucwziE/eZql05dZUBQu+yRQKtueMxJauKwysHbiXj6F2wsNXUumXACFzAeRx9D/6/V pBZUZWAYs4p13o4grqh0QJIBv3QBdimwPuaapizMbOKJVzsGAB8X6SIjUeFu3aa02Ge6 VCLA== X-Forwarded-Encrypted: i=1; AJvYcCVLyZkmkglgAFl2pFb68ySJDYwZRnl97C6j58gtakvlGkLT60oNIV4I/++ROzYDx7Ycxr09vdySPYcUyTw=@vger.kernel.org X-Gm-Message-State: AOJu0Yz8/wP5/dpfc0BAzoM9pNLMmjiShjIhnqL6B/UWOx/hFZPHyn9H YHXIu054O9vrLby5FsvxpH3NIbViN+XTUF8dFodTgC588btphKoTBdaLMXUx6omoB+b+Q6S6ITi SC8+FJQ== X-Google-Smtp-Source: AGHT+IEwizCIlZ2mc7buHdqoiSSOfJ7BZ84a8AqmrE27Kj7Vd7q1ky5apXKAZA8h3DcPgWfNdVvfYIQ7kII8 X-Received: from pgbdo14.prod.google.com ([2002:a05:6a02:e8e:b0:af2:48c3:360f]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:a92:b0:1f5:6a1a:329b with SMTP id adf61e73a8af0-1fe4300f787mr38309635637.32.1742947582873; Tue, 25 Mar 2025 17:06:22 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:07 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-10-irogers@google.com> Subject: [PATCH v3 09/35] perf vendor events: Update clearwaterforest events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic of OCR.DEMAND_DATA_RD.ANY_RESPONSE and OCR.DEMAND_RFO.ANY_RESPONSE to be cache. Signed-off-by: Ian Rogers --- .../arch/x86/clearwaterforest/cache.json | 20 +++++++++++++++++ .../arch/x86/clearwaterforest/other.json | 22 ------------------- 2 files changed, 20 insertions(+), 22 deletions(-) delete mode 100644 tools/perf/pmu-events/arch/x86/clearwaterforest/other.j= son diff --git a/tools/perf/pmu-events/arch/x86/clearwaterforest/cache.json b/t= ools/perf/pmu-events/arch/x86/clearwaterforest/cache.json index 875361b30f1d..17f8bfba56bc 100644 --- a/tools/perf/pmu-events/arch/x86/clearwaterforest/cache.json +++ b/tools/perf/pmu-events/arch/x86/clearwaterforest/cache.json @@ -140,5 +140,25 @@ "EventName": "MEM_UOPS_RETIRED.STORE_LATENCY", "SampleAfterValue": "1000003", "UMask": "0x6" + }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" } ] diff --git a/tools/perf/pmu-events/arch/x86/clearwaterforest/other.json b/t= ools/perf/pmu-events/arch/x86/clearwaterforest/other.json deleted file mode 100644 index 80454e497f83..000000000000 --- a/tools/perf/pmu-events/arch/x86/clearwaterforest/other.json +++ /dev/null @@ -1,22 +0,0 @@ -[ - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - } -] --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B92EE335C7 for ; Wed, 26 Mar 2025 00:06:32 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947595; cv=none; b=Jbwqq6At4HXmkVuDujXAs1CX5RbxfPYAWv/k89bzyLoU3Gg34OnKUfJ8A+q/agfTF5VOmelxSwrSmScKgYgrD5q2s+tJS8V/hboAoL5YyKJ4PWpOLJ5o5wvHkdzEDOExU2PU0Du+HaKIGSdVloGjQXvahvmIu7CYwbbibw5pNKw= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947595; c=relaxed/simple; bh=p7Mb6vIPPHoTDt3eSFoj41xhRJUybxfZh++R1hxEinA=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=kGQZqZ65VjezC69utuq9ENDLO3ZDHNowzyii/FYGkrOcyg3DsERx+WE9DQYJTdj8OusNdav4LnkazGUzOa3xg/E0IoMSCtT1K2LdTIINxR1ri8bk/rufdB+nMnMZd7hIjVESd2GBDN6OXzfuHrArmdV+D7TzzGTVADT+bXcNRmk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=nzYudbZo; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="nzYudbZo" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2242ade807fso174342275ad.2 for ; Tue, 25 Mar 2025 17:06:32 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947592; x=1743552392; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=0YeSs3HV21hPqxME9Y/uPxwXurItuHzesMa147bUg4c=; b=nzYudbZo+7+41sr4otsvju8B14S3erTbwI0jzbqOjk0VgixGpAqsBLAWIjAbRwxiWe C9jfbtCr0nvVyZjxUGSK63TZyYIj24sduzoOam861G4dDIV4335CEvFR9eJykabG8Zpp IuknpYOYxSKtZvmagkOMuUMUKO8OT6ecZ6pDDpXB90wrg4XzFS+1HoA3xBMceFk7NsCW KlMwFm86Yd4z5s+lsq9acOSAFia6Sz+ltwfeUdcCfz5fVoNOUoy9Vex8x53yedElRu9t uTxY+muEqRPZEobwGZlXqTXBScwHZqSpqM+q2w5/bqLsPtnqcnQeQNFK4LAuXDG8SfU3 EtPQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947592; x=1743552392; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=0YeSs3HV21hPqxME9Y/uPxwXurItuHzesMa147bUg4c=; b=JMk9kY+eSRMjqdgATtHnH7NqdrGu65yD/nJ1hJkL4/2S+yxBowELcDA4S8IvbwkMuF +7zl8KGgjIATwx/pHhXu9BFDqUVRaoLbk1izD4VOzlsu45ScqlysKenP0TSWy4Iw3qSk 2ekSpnvX+gt4g7ckyFHol16805PMB1F5URHLRuSc30+d8VSF8U2Mt72Oo5/1+WXzr2B7 R1/9lGH+WkVqKE/927PgW0vMFrKFsPViLAAuXyQJL8u4mry64MADftlvn3R1oPu5/3di QS4iqkflcOB5hbPZTrGRohFWPxsHO1AhpTNQfad4Tw8Iz7s3HVxyvGnsScITSJTlSt10 CCNQ== X-Forwarded-Encrypted: i=1; AJvYcCXr5kFvVLXy612hnx2oSferxsoycSnU/4PJvbblO65G/2s5LlMSYR57g1aKz4lj/YpJTCKHfNSEAonyzEM=@vger.kernel.org X-Gm-Message-State: AOJu0Yz+GRrLC7xg88uZ0viErnvZMx+A3WhuN1wMzNBWe3UPl1q3PsHr 2NB+BsFFaSGDCG7OiZSz+fIggAwnCttRUzTVPd5sg6Y5j7sO5X2lwrNq5XiTElPUrdvSEiip0Yk MZpm/IQ== X-Google-Smtp-Source: AGHT+IG6AedIVGmIkw8bKTrSUrnnbrgYpGHeGCz32LZ/HhCMwXR6yo24H4J3wXezct2suzxMRvhzbMEV+8OI X-Received: from pjbmf11.prod.google.com ([2002:a17:90b:184b:b0:2ef:d283:5089]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:903:41c8:b0:223:5ace:eccf with SMTP id d9443c01a7336-22780d8a9e9mr316100065ad.25.1742947592108; Tue, 25 Mar 2025 17:06:32 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:08 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-11-irogers@google.com> Subject: [PATCH v3 10/35] perf vendor events: Update elkhartlake events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and memory. Signed-off-by: Ian Rogers --- .../arch/x86/elkhartlake/cache.json | 192 +++++++++ .../arch/x86/elkhartlake/memory.json | 202 +++++++++ .../arch/x86/elkhartlake/other.json | 394 ------------------ 3 files changed, 394 insertions(+), 394 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/elkhartlake/cache.json b/tools/= perf/pmu-events/arch/x86/elkhartlake/cache.json index 7882dca9d5e1..1bb42acf1d48 100644 --- a/tools/perf/pmu-events/arch/x86/elkhartlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/elkhartlake/cache.json @@ -357,6 +357,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.COREWB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -367,6 +377,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have an outstanding request. Returns the number of cycles unt= il the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.COREWB_M.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8003000000000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -427,6 +457,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -487,6 +527,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve an outstanding request. Returns the number of cycles until the response = is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.L3_HIT", "Counter": "0,1,2,3", @@ -553,6 +614,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache.", "Counter": "0,1,2,3", @@ -613,6 +695,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have an outstan= ding request. Returns the number of cycles until the response is received (= i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -623,6 +715,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that have any type of resp= onse.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that were supplied by the = L3 cache where a snoop was sent, the snoop hit, and modified data was forwa= rded.", "Counter": "0,1,2,3", @@ -633,6 +735,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -693,6 +805,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have an outstanding request. Returns th= e number of cycles until the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -753,6 +885,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -813,6 +955,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have an outstanding request. Returns the numb= er of cycles until the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.L1WB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -823,6 +985,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.L2WB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x2000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -843,6 +1015,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -903,6 +1085,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e an outstanding request. Returns the number of cycles until the response i= s received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that were supplied by= the L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/elkhartlake/memory.json b/tools= /perf/pmu-events/arch/x86/elkhartlake/memory.json index 34306ec24e9b..260a488540bb 100644 --- a/tools/perf/pmu-events/arch/x86/elkhartlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/elkhartlake/memory.json @@ -25,6 +25,16 @@ "SampleAfterValue": "200003", "UMask": "0x4" }, + { + "BriefDescription": "Counts all code reads that were supplied by D= RAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.ALL_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000044", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all code reads that were not supplied = by the L3 cache.", "Counter": "0,1,2,3", @@ -45,6 +55,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all code reads that were supplied by D= RAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.ALL_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000044", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -65,6 +85,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -85,6 +115,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -105,6 +155,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.L3_MISS", "Counter": "0,1,2,3", @@ -127,6 +198,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3", @@ -147,6 +239,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -167,6 +269,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -187,6 +299,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -207,6 +339,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -227,6 +379,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -317,6 +479,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -337,6 +509,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that were not supplie= d by the L3 cache.", "Counter": "0,1,2,3", @@ -357,6 +539,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.UC_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100184000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts uncached memory reads that were not su= pplied by the L3 cache.", "Counter": "0,1,2,3", @@ -377,6 +569,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.UC_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100184000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts uncached memory writes that were not s= upplied by the L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/elkhartlake/other.json b/tools/= perf/pmu-events/arch/x86/elkhartlake/other.json index 57613207f7ad..35cdbfa617e7 100644 --- a/tools/perf/pmu-events/arch/x86/elkhartlake/other.json +++ b/tools/perf/pmu-events/arch/x86/elkhartlake/other.json @@ -116,26 +116,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all code reads that were supplied by D= RAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.ALL_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000044", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all code reads that were supplied by D= RAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.ALL_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000044", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts all code reads that have an outstandin= g request. Returns the number of cycles until the response is received (i.e= . XQ to XQ latency).", "Counter": "0,1,2,3", @@ -146,180 +126,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.COREWB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have an outstanding request. Returns the number of cycles unt= il the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.COREWB_M.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8003000000000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve an outstanding request. Returns the number of cycles until the response = is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have an outstan= ding request. Returns the number of cycles until the response is received (= i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3", @@ -330,146 +136,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that have any type of resp= onse.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have an outstanding request. Returns th= e number of cycles until the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have an outstanding request. Returns the numb= er of cycles until the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.L1WB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.L2WB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x2000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts miscellaneous requests, such as I/O ac= cesses, that have any type of response.", "Counter": "0,1,2,3", @@ -500,46 +166,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e an outstanding request. Returns the number of cycles until the response i= s received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -560,26 +186,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.UC_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100184000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.UC_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100184000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts uncached memory reads that have an out= standing request. Returns the number of cycles until the response is receiv= ed (i.e. XQ to XQ latency).", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oa1-f74.google.com (mail-oa1-f74.google.com [209.85.160.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 0CD0F3D6A for ; Wed, 26 Mar 2025 00:06:41 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.160.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947611; cv=none; b=eHxXoE/aNqqn42doeZ2fAoFAe59EoVHKrU0O78L/hKXONfWknUWGxUut6+D3DwcmiDnhN3jsqClx3YhY/h2kw+ZXR9Y1qVRwNc4MQosIvKyrVUvp8CJCBLiYrifo8/1dC8MSy9fu8fh+gyef15WJRl/QXuviO2stTYkkIqkIV10= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947611; c=relaxed/simple; bh=nb+8fLuWTinOKP0A7sCwarDMUI5I2Cc/HjNA02PyftU=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=P0F1CFkaNYx7ZVdpJX7sfAaWeoad/867rmOeh5nLzv6HLSPsGMExX6BTIBhkAdOOqEHA1Tyw5QydAhE3h/G5LrtLdVuCFJLG+04WT29zf6Qvd8A2GxJkJwxqEbqJhF5Uy/0R7VhmcsdDs6rX8OFR6bJ+kWwgw0odBl0N5cdzJw0= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=zq2nBI/B; arc=none smtp.client-ip=209.85.160.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="zq2nBI/B" Received: by mail-oa1-f74.google.com with SMTP id 586e51a60fabf-2c268416ff5so1566220fac.3 for ; Tue, 25 Mar 2025 17:06:41 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947601; x=1743552401; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=o8pJbBgougiXAgYO3m/+fYfim8xB1NDfiBeZe3FEsv0=; b=zq2nBI/BbJlUdYy0S0qihDB+klsNmfqx2CaVFw/SIO9Wha/p0uEKE11X39y6JSAXcv uruvBMPlD1cZAgyTQVROjRR6IkH4TEIXhoEph4PT/haWgcSRZjw7qUg9bk+KfrtJFeJa 2TDopL7ken8mwvOAbFBKquJofiWV4JHzqKflMdvrIokGfhH53kv587dHsHclr+fP7Wzc Rhj05oROUQF56QnYSeJbXWRAcRguqjrNl6p99ixNUijG5WNIue4AhJWR796hCxdLGDov dGm5fl2YfgLN6F1o3WcViTA3uN8uEPqlY1O5qaBvABqWT+uWVomrfcz4qM0y8f2sqihe UbKQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947601; x=1743552401; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=o8pJbBgougiXAgYO3m/+fYfim8xB1NDfiBeZe3FEsv0=; b=xFQD1WmfEz4w9acoYjhI/oG1HZ5ruCi4Eqb9h2IdjROV9wj+E8n9dTPMH+BC/P4OrF aIZS4fROQA1gGMWsvySMpn8usUV549R3Fj1jgeG2+qrzQneXni0Xl/TyD4axe1dYkrPh 01+beAj4UncZwBL5gWKbPs0WBMqwA+UIhIhLvB04Lp6c69Qez/SwgYvLaB7omXOIyQto eHKLuZn6DM78KSsXchqLyoabHivSzOVpMfqAzVV2obFLn69xkfbUdfMunQ5qC4N6iumL ZM7ohdtE1C25US18bx34rF3+cXtAfNJAp+0nNAwJtYg9bA/x9oLDO4G/DuehZ0vPFITo wv6g== X-Forwarded-Encrypted: i=1; AJvYcCXp2m3pCdYPLULLAGA3AZ1ztQJYCR2sQ5OIj9RCwAnMAbjesMFM5qebm4chzIgksVhSGXIZr6TG5Rg1IFw=@vger.kernel.org X-Gm-Message-State: AOJu0YzIOCsWVI2m77E8NDDE6QgS6QMExLc7+KjqxUiDfQWm8xh/9g5O yPziAAgqJpAKOwyZMrZG79eYO0kR1FDudYaLC3cZz+wLUCYcL2SbH9rF12lW+a+fvI2NzoXYfvw xsDr5bA== X-Google-Smtp-Source: AGHT+IHj7Sl2ifpOOYRlochK2dqNCXktzQ19Bvga+5rOOE+SyI24ImiZhG8FPkealCjRDpJELtMJttHpEdi/ X-Received: from oabwx5.prod.google.com ([2002:a05:6870:7e05:b0:2b3:580c:19f4]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6870:6e84:b0:2bd:456c:92a with SMTP id 586e51a60fabf-2c7802981d4mr10887354fac.10.1742947600918; Tue, 25 Mar 2025 17:06:40 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:09 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-12-irogers@google.com> Subject: [PATCH v3 11/35] perf vendor events: Update emeraldrapids events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/emeraldrapids/cache.json | 100 ++++ .../arch/x86/emeraldrapids/emr-metrics.json | 471 +++++++++--------- .../arch/x86/emeraldrapids/memory.json | 170 +++++++ .../arch/x86/emeraldrapids/other.json | 328 ------------ .../arch/x86/emeraldrapids/pipeline.json | 58 +++ 5 files changed, 563 insertions(+), 564 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/emeraldrapids/cache.json b/tool= s/perf/pmu-events/arch/x86/emeraldrapids/cache.json index 3b0581151d63..0b2b36a30075 100644 --- a/tools/perf/pmu-events/arch/x86/emeraldrapids/cache.json +++ b/tools/perf/pmu-events/arch/x86/emeraldrapids/cache.json @@ -569,6 +569,16 @@ "SampleAfterValue": "1000003", "UMask": "0x3" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit in the L3 or were snooped from another co= re's caches on the same socket.", "Counter": "0,1,2,3", @@ -609,6 +619,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit in the L3 o= r were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -689,6 +709,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC0002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit in= the L3 or were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -729,6 +759,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts data load hardware prefetch requests t= o the L1 data cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L1D.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetches (which bring data = to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L2.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10070", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L3.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x12380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t hit in the L3 or were snooped from another core's caches on the same sock= et.", "Counter": "0,1,2,3", @@ -739,6 +799,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L3.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x90002380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10808", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC4477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit in the L3 or were snooped from another core's caches on the sa= me socket.", "Counter": "0,1,2,3", @@ -779,6 +869,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F33004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by a cache on a remote socket where a snoop was sent= and data was returned (Modified or Not Modified).", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/emeraldrapids/emr-metrics.json = b/tools/perf/pmu-events/arch/x86/emeraldrapids/emr-metrics.json index d3b51fa6ec1c..1c4301ca9892 100644 --- a/tools/perf/pmu-events/arch/x86/emeraldrapids/emr-metrics.json +++ b/tools/perf/pmu-events/arch/x86/emeraldrapids/emr-metrics.json @@ -300,7 +300,7 @@ "ScaleUnit": "1per_instr" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6) / (5 * tma_info_core_co= re_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -312,7 +312,7 @@ "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma= _core_bound_group", "MetricName": "tma_amx_busy", - "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tm= a_backend_bound > 0.2", + "MetricThreshold": "tma_amx_busy > 0.5 & (tma_core_bound > 0.1 & t= ma_backend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -320,12 +320,12 @@ "MetricExpr": "78 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -335,7 +335,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_inf= o_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -351,12 +351,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -371,7 +371,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -379,7 +379,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_= bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_d= ram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_st= ore_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_load= s + tma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_= l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_s= tore_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sha= ring + tma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_mem= ory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound = + tma_dram_bound + tma_store_bound)) * (tma_store_latency / (tma_store_late= ncy + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtl= b_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependen= cy + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_= bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma= _l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_fb= _full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tm= a_store_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tm= a_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_= latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_s= tore_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharin= g + tma_split_stores + tma_store_latency + tma_streaming_stores)) + tma_mem= ory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store= + tma_false_sharing + tma_split_stores + tma_store_latency + tma_streaming= _stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -387,22 +387,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_amx_busy + tma_ports_utilization) + tma_c= ore_bound * tma_amx_busy / (tma_divider + tma_serializing_operation + tma_a= mx_busy + tma_ports_utilization) + tma_core_bound * (tma_ports_utilization = / (tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utili= zation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_amx_busy= + tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_c= ore_bound * tma_amx_busy / (tma_amx_busy + tma_divider + tma_ports_utilizat= ion + tma_serializing_operation) + tma_core_bound * (tma_ports_utilization = / (tma_amx_busy + tma_divider + tma_ports_utilization + tma_serializing_ope= ration)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fe= tch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers= + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredict= s) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches= )) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_sw= itches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_= mite + tma_dsb + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetc= h_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers += tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts)= / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_branches))= / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_m= isses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms / (tma_ds= b + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_b= ranch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_othe= r_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_resteers + tma_c= lears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_itlb_mis= ses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) += tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ms)) + 10 * tma_m= icrocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_b= ranch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes = + tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_inf= o_thread_clks * tma_ports_utilized_0) / (tma_divider + tma_serializing_oper= ation + tma_amx_busy + tma_ports_utilization) + tma_microcode_sequencer / (= tma_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_m= icrocode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_switches + tma_bra= nch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_other_= mispredicts / tma_branch_mispredicts) / (tma_clears_resteers + tma_mispredi= cts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_dsb_swit= ches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) + t= ma_fetch_bandwidth * tma_ms / (tma_dsb + tma_mite + tma_ms)) + 10 * tma_mic= rocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_bra= nch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes + = tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_= thread_clks * tma_ports_utilized_0) / (tma_amx_busy + tma_divider + tma_por= ts_utilization + tma_serializing_operation) + tma_microcode_sequencer / (tm= a_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_mic= rocode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -410,7 +410,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound += tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_store_latency + tma_false_sharing + tma_split_stores + tma_= streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dram_bound= + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tma_sto= re_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -418,7 +418,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency + tma_streaming_stores - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -426,7 +426,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -438,10 +438,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -450,7 +450,7 @@ { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tm= a_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;Default;TmaL2;TopdownL2= ;tma_L2_group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -463,24 +463,24 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -488,8 +488,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRE= D.MS_FLOWS", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -497,24 +497,24 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -522,7 +522,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -530,32 +530,32 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((81 * tma_info_system_core_frequency - 4.4 * tma_i= nfo_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (OCR.DEMAN= D_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) + (79 * tma_info_system_core_fre= quency - 4.4 * tma_info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XS= NP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / t= ma_info_thread_clks", + "MetricExpr": "(76.6 * tma_info_system_core_frequency * (MEM_LOAD_= L3_HIT_RETIRED.XSNP_FWD * (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMA= ND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD= ))) + 74.6 * tma_info_system_core_frequency * MEM_LOAD_L3_HIT_RETIRED.XSNP_= MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_= info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -566,24 +566,24 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "(79 * tma_info_system_core_frequency - 4.4 * tma_in= fo_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD= _L3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR= .DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WIT= H_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / t= ma_info_thread_clks", + "MetricExpr": "74.6 * tma_info_system_core_frequency * (MEM_LOAD_L= 3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEM= AND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR= .DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT= / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -592,8 +592,8 @@ "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%" }, { @@ -601,7 +601,7 @@ "MetricExpr": "MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_cl= ks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -611,7 +611,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -619,34 +619,34 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEM= ORY_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMOR= Y_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", - "MetricExpr": "(170 * tma_info_system_core_frequency * cpu@OCR.DEM= AND_RFO.L3_MISS\\,offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core= _frequency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", + "MetricExpr": "(170 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_MISS@offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core_frequ= ency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, @@ -667,7 +667,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -678,7 +678,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -696,7 +696,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -705,15 +705,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -721,8 +721,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED2.SCALAR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -730,8 +730,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIR= ED2.VECTOR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6= , tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -739,8 +739,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -748,8 +748,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -757,8 +757,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_= 2", "ScaleUnit": "100%" }, { @@ -769,27 +769,27 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%" }, { @@ -797,8 +797,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -809,28 +809,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -858,7 +858,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -866,7 +866,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -874,10 +874,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -938,11 +939,11 @@ "MetricExpr": "(FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.P= ORT_1 + FP_ARITH_DISPATCHED.PORT_5) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -955,20 +956,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask= \\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask= \\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -1005,13 +1006,13 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node" + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -1029,7 +1030,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -1037,7 +1038,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -1045,7 +1046,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -1053,7 +1054,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -1061,7 +1062,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Pr= ecision instruction (lower number means higher occurrence rate)", @@ -1069,7 +1070,7 @@ "MetricGroup": "Flops;FpScalar;InsType;Server", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_hp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -1077,7 +1078,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1132,7 +1133,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 13", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1269,7 +1270,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1334,21 +1335,21 @@ "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system= _time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_dram_bw", - "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW" + "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW." }, { "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R= 2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_sys= tem_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", - "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW" + "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW." }, { "BriefDescription": "Average Off-core access BW for Reads-to-Core = (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_inf= o_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", - "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches" + "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches." }, { "BriefDescription": "STLB (2nd level TLB) code speculative misses = per kilo instruction (misses of any page-size that complete the page walk)", @@ -1376,8 +1377,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1398,18 +1399,18 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D0x1@", + "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" @@ -1472,14 +1473,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1490,7 +1490,7 @@ }, { "BriefDescription": "Average latency of data read request to exter= nal DRAM memory [in nanoseconds]", - "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / cha_0@event\\=3D0x0@", + "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / uncore_cha_0@event\\=3D0x1@", "MetricGroup": "MemOffcore;MemoryLat;Server;SoC", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data= -read prefetches" @@ -1500,11 +1500,11 @@ "MetricExpr": "UNC_CHA_RxC_IRQ1_REJECT.PA_MATCH / UNC_CHA_CLOCKTIC= KS", "MetricGroup": "LockCont;MemOffcore;Server;SoC", "MetricName": "tma_info_system_mem_irq_duplicate_address", - "MetricThreshold": "(tma_info_system_mem_irq_duplicate_address > 0= .1)" + "MetricThreshold": "tma_info_system_mem_irq_duplicate_address > 0.= 1" }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1538,7 +1538,7 @@ }, { "BriefDescription": "Socket actual clocks when any core is active = on that socket", - "MetricExpr": "cha_0@event\\=3D0x0@", + "MetricExpr": "uncore_cha_0@event\\=3D0x1@", "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, @@ -1568,7 +1568,7 @@ "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1577,15 +1577,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1595,13 +1594,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1617,14 +1616,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 9" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1633,7 +1632,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%" }, { @@ -1641,8 +1640,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1650,8 +1649,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 = + INT_VEC_RETIRED.VNNI_256) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1659,8 +1658,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1668,7 +1667,7 @@ "MetricExpr": "max((EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.= STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1677,7 +1676,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1686,7 +1685,7 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.= STALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1695,7 +1694,7 @@ "MetricExpr": "4.4 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1704,17 +1703,17 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.S= TALLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(37 * tma_info_system_core_frequency - 4.4 * tma_in= fo_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETIRE= D.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "32.6 * tma_info_system_core_frequency * (MEM_LOAD_R= ETIRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2= )) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1722,19 +1721,19 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "DefaultMetricgroupName": "TopdownL2", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1751,7 +1750,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1759,39 +1758,39 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from local memory", - "MetricExpr": "(109 * tma_info_system_core_frequency - 37 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * (1 + MEM_= LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "72 * tma_info_system_core_frequency * MEM_LOAD_L3_M= ISS_RETIRED.LOCAL_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1= _MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1800,7 +1799,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1816,19 +1815,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling)", + "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling).", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma= _mem_bandwidth_group", "MetricName": "tma_mba_stalls", - "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2= & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mba_stalls > 0.1 & (tma_mem_bandwidth > 0.= 2 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0= .2)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1837,32 +1836,32 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Backend;Default;TmaL2;TopdownL2;tma_L2_group;tma_b= ackend_bound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / (tma_r= etiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1883,7 +1882,7 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * INT_= MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1897,17 +1896,17 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clk= s / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clks = / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1915,11 +1914,11 @@ }, { "BriefDescription": "This metric estimates the fraction of cycles = when the CPU was stalled due to switches of uop delivery to the Microcode S= equencer (MS)", - "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1\\,edge\\=3D= 0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", + "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D1\\,edge@ / (UO= PS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tm= a_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_mac= hine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_o= peration", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, { @@ -1928,7 +1927,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1936,7 +1935,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -1950,19 +1949,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1971,7 +1970,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%" }, { @@ -1980,7 +1979,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_po= rts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b,= tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1989,7 +1988,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector= _512b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_5, tm= a_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1998,7 +1997,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= ts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, = tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector= _256b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2006,8 +2005,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL)) / tm= a_info_thread_clks if ARITH.DIV_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - EXE_= ACTIVITY.BOUND_ON_LOADS else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EX= E_ACTIVITY.2_3_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -2015,8 +2014,8 @@ "MetricExpr": "(EXE_ACTIVITY.EXE_BOUND_0_PORTS + max(RS.EMPTY_RESO= URCE - RESOURCE_STALLS.SCOREBOARD, 0)) / tma_info_thread_clks * (CYCLE_ACTI= VITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -2024,7 +2023,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -2034,8 +2033,8 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%" }, { @@ -2044,32 +2043,32 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote cache in other socket= s including synchronizations issues", - "MetricExpr": "((170 * tma_info_system_core_frequency - 37 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + (170 * = tma_info_system_core_frequency - 37 * tma_info_system_core_frequency) * MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD= _RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(133 * tma_info_system_core_frequency * MEM_LOAD_L3= _MISS_RETIRED.REMOTE_HITM + 133 * tma_info_system_core_frequency * MEM_LOAD= _L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETI= RED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote memory", - "MetricExpr": "(190 * tma_info_system_core_frequency - 37 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "153 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2082,7 +2081,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks += tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -2091,8 +2090,8 @@ "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%" }, { @@ -2101,7 +2100,7 @@ "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%" }, @@ -2111,7 +2110,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -2119,8 +2118,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -2128,7 +2127,7 @@ "MetricExpr": "(XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS) / tma_in= fo_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -2137,8 +2136,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -2146,8 +2145,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -2155,8 +2154,8 @@ "MetricExpr": "(MEM_STORE_RETIRED.L2_HIT * 10 * (1 - MEM_INST_RETI= RED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_= LOADS / MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE= _REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -2173,7 +2172,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -2181,31 +2180,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2213,7 +2212,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -2222,7 +2221,7 @@ "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_cl= ks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%" }, @@ -2231,8 +2230,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/emeraldrapids/memory.json b/too= ls/perf/pmu-events/arch/x86/emeraldrapids/memory.json index 41d4120d4dae..981e573330cd 100644 --- a/tools/perf/pmu-events/arch/x86/emeraldrapids/memory.json +++ b/tools/perf/pmu-events/arch/x86/emeraldrapids/memory.json @@ -173,6 +173,16 @@ "SampleAfterValue": "1000003", "UMask": "0x2" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the local socket's L1, L= 2, or L3 caches.", "Counter": "0,1,2,3", @@ -183,6 +193,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -193,6 +233,46 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were n= ot supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -203,6 +283,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t missed the local socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -223,6 +323,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -253,6 +363,56 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x70C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x733004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that missed the local= socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -273,6 +433,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.WRITE_ESTIMATE.MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0xFBFF80822", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/emeraldrapids/other.json b/tool= s/perf/pmu-events/arch/x86/emeraldrapids/other.json index c424facf1b95..df4019ff7883 100644 --- a/tools/perf/pmu-events/arch/x86/emeraldrapids/other.json +++ b/tools/perf/pmu-events/arch/x86/emeraldrapids/other.json @@ -7,274 +7,6 @@ "SampleAfterValue": "1000003", "UMask": "0x8" }, - { - "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb7", - "EventName": "EXE.AMX_BUSY", - "SampleAfterValue": "2000003", - "UMask": "0x2" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC0002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts data load hardware prefetch requests t= o the L1 data cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L1D.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches (which bring data = to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L2.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10070", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L3.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x12380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L3.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x90002380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10808", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC4477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x70C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F33004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x733004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -285,66 +17,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.WRITE_ESTIMATE.MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0xFBFF80822", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "Deprecated": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.COUNT", - "Invert": "1", - "SampleAfterValue": "100003", - "UMask": "0x7" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", - "Counter": "0,1,2,3,4,5,6,7", - "Deprecated": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.CYCLES", - "SampleAfterValue": "1000003", - "UMask": "0x7" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/emeraldrapids/pipeline.json b/t= ools/perf/pmu-events/arch/x86/emeraldrapids/pipeline.json index 50cacfbbc7cf..c16b63979c55 100644 --- a/tools/perf/pmu-events/arch/x86/emeraldrapids/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/emeraldrapids/pipeline.json @@ -367,6 +367,14 @@ "SampleAfterValue": "1000003", "UMask": "0x4" }, + { + "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb7", + "EventName": "EXE.AMX_BUSY", + "SampleAfterValue": "2000003", + "UMask": "0x2" + }, { "BriefDescription": "Cycles total of 1 uop is executed on all port= s and Reservation Station was not empty.", "Counter": "0,1,2,3,4,5,6,7", @@ -740,6 +748,56 @@ "SampleAfterValue": "100003", "UMask": "0x2" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7" + }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "Deprecated": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.COUNT", + "Invert": "1", + "SampleAfterValue": "100003", + "UMask": "0x7" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", + "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.CYCLES", + "SampleAfterValue": "1000003", + "UMask": "0x7" + }, { "BriefDescription": "TMA slots where no uops were being issued due= to lack of back-end resources.", "Counter": "0,1,2,3,4,5,6,7", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id A60C0433B1 for ; Wed, 26 Mar 2025 00:06:50 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947614; cv=none; b=dYge2sXxjv8Rl6PaoO5mZCTL8l6gdpOcBctIZqB6hBxqSTAqi/pfDLXhdQrXMDACQcUAqncfxxTykRNQlXrvFCOjBT2XzYn5bZloEiUjrN0KVWBeVRBpUhstYxTh0lYTALW6AtyytYgVi6Um8pynbkJTzL3T7BmOgvz+t6I5teU= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947614; c=relaxed/simple; bh=Jcm7UL8/lFDhQIlPhuAgLeKF3UATwKiIBixq/P6AT10=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=EEy1jf2bVS0Yql9+sEI8dqY5jZRxKob9IHvJg2j/oTT4ZE2Fi9iH00d8jD/tXjLXlnC8HWgYqQsuAeZaN15eTeScSM1cRVw7zDGG3z+GlLUuXmTeDUI1yUsBtsyKeu2BgtcxGDT4RG2jL0TvqM2iz9spfCeMyXjTXnANr9tbqVs= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=V7Ps2VjA; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="V7Ps2VjA" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff58318acaso17412432a91.0 for ; Tue, 25 Mar 2025 17:06:50 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947610; x=1743552410; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=9YC3LRLM5jbvcC6M7msVEOfa40sR0TPKU7PhwtgnPbI=; b=V7Ps2VjAZGtoa0tNEZKL4zVsi404kGbdH03eAHVOK1/SyD2jO94vOf03HlpPrXOwBg 1ccsus0NM1YapFt6lqlAbbeN3DqZs2r9EDIx5ukccdXjA+j6bpshIUOPCMMO62S4Ux9t wmE7o2tVLJP+ZHCAgbfIsQFmqmexgk6yuhyZfzdful4ACMy4VdD73JswBIhNkM+H9ql1 SjdcKd6BwdOSqZQNuTCN9y6xFnD2xW/by4RYkPfy8VeDRutpE+xiZKtjZedm4e0VvFRx djRm/G9A/H+uPpID01RhmWon7bed9Zsg5hfYdCBCBI3fUuPjHLyFyIC1kZR5YCDiO1Su aJhQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947610; x=1743552410; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=9YC3LRLM5jbvcC6M7msVEOfa40sR0TPKU7PhwtgnPbI=; b=FA9H0ZRX8kdtsSx4nxdrNKFzuVMUMqdj4yEmBfO+slXcbBnE4RY9oR/lcBZfneCbjB QSSeRC1IOAYOseafzGsdEs9k22MszVsLQmqwP2PRJEmRXIaOMVuxaJ5SI319Uol0Ut8X sZmjZqJl4fh6yvwHGXC1jQcjq/uawdP3mbxj56luGh7vJv2/A2M9qHSj+FoKZ9mDItpM +MksKpQZc3tvhGwFPT8KnWVlVx9fqkaiXkz3GojMaC+JGACw9eCKjeCHhoj6jKzZKYUD IXTEMKHlkzs+mYxKXsrwWlIp2h6LOcq1P3LDgo8lKzszFQYisKngXCm6peNSiBdJyF6q 5+/Q== X-Forwarded-Encrypted: i=1; AJvYcCUnqADRBE7vewnuKDNemrqaKbvtZbHI9rZh1Pbgt09rBolsapWf/e8WMKznTmhrMJ8FaWlqeIBENV/ouJ0=@vger.kernel.org X-Gm-Message-State: AOJu0Yy040bOyRjFItWuLh7XLP6TJ7RxFPrsvfFoiOMOotiPoy1Ixps0 ANwCdoMHSRoWFlF7f/4Zi5TltXBwaLZ8GiOhHhJHfQ8n+oXvVD1pEcoFe6rhlkcoQ9uI9SECRec a+kBgKA== X-Google-Smtp-Source: AGHT+IGgWvs255vvl2P+Nxt4qKQDdAMoSG1zV6Eb9WJybL2r/TPEJPlhJowV59o6GngjeZHvJPgFxITujW7Z X-Received: from pgbcz8.prod.google.com ([2002:a05:6a02:2308:b0:ae6:172:38f3]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:6da7:b0:1f5:8eec:e516 with SMTP id adf61e73a8af0-1fe4330009fmr38755715637.32.1742947609933; Tue, 25 Mar 2025 17:06:49 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:10 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-13-irogers@google.com> Subject: [PATCH v3 12/35] perf vendor events: Update grandridge events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.05 to v1.07. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/grandridge/cache.json | 150 ++++- .../arch/x86/grandridge/counter.json | 2 +- .../arch/x86/grandridge/frontend.json | 8 + .../arch/x86/grandridge/grr-metrics.json | 521 +----------------- .../pmu-events/arch/x86/grandridge/other.json | 28 - .../arch/x86/grandridge/pipeline.json | 51 +- .../arch/x86/grandridge/uncore-cache.json | 45 +- .../arch/x86/grandridge/uncore-memory.json | 338 ++++++++++++ tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- 9 files changed, 582 insertions(+), 563 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/grandridge/cache.json b/tools/p= erf/pmu-events/arch/x86/grandridge/cache.json index 04802e254e51..21671c65d6dd 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/cache.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/cache.json @@ -1,4 +1,91 @@ [ + { + "BriefDescription": "Counts the number of L1D cacheline (dirty) ev= ictions caused by load misses, stores, and prefetches.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x51", + "EventName": "DL1.DIRTY_EVICTION", + "PublicDescription": "Counts the number of L1D cacheline (dirty) e= victions caused by load misses, stores, and prefetches. Does not count evi= ctions or dirty writebacks caused by snoops. Does not count a replacement = unless a (dirty) line was written back.", + "SampleAfterValue": "200003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Exclusive state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.E", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Exclusive state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x4" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Forward state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.F", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Forward state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x10" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Modified state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.M", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Modified state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x8" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Shared state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.S", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Shared state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x2" + }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = evicted due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.NON_SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= evicted due to an L2 cache fill. Increments on the core that brought the l= ine in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x2" + }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = silently dropped due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= silently dropped due to an L2 cache fill. Increments on the core that bro= ught the line in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that r= esulted in a Hit from a front door request only (does not include rejects o= r recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.HIT", + "SampleAfterValue": "200003", + "UMask": "0x2" + }, + { + "BriefDescription": "Counts the number of total L2 Cache Accesses = that resulted in a Miss from a front door request only (does not include re= jects or recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.MISS", + "SampleAfterValue": "200003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that m= iss the L2 and get BBL reject short and long rejects, per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.REJECTS", + "SampleAfterValue": "200003", + "UMask": "0x4" + }, { "BriefDescription": "Counts the number of cacheable memory request= s that miss in the LLC. Counts on a per core basis.", "Counter": "0,1,2,3,4,5,6,7", @@ -35,7 +122,7 @@ "UMask": "0x1" }, { - "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to an icache or itlb miss which hit in the LLC.", + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to an ICACHE or ITLB miss which hit in the LLC. If the= core has access to an L3 cache, an LLC hit refers to an L3 cache hit, othe= rwise it counts zeros.", "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0x35", "EventName": "MEM_BOUND_STALLS_IFETCH.LLC_HIT", @@ -43,7 +130,7 @@ "UMask": "0x6" }, { - "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to an icache or itlb miss which missed all the caches.= ", + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to an ICACHE or ITLB miss which missed all the caches.= If the core has access to an L3 cache, an LLC miss refers to an L3 cache m= iss, otherwise it is an L2 cache miss.", "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0x35", "EventName": "MEM_BOUND_STALLS_IFETCH.LLC_MISS", @@ -68,7 +155,7 @@ "UMask": "0x1" }, { - "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to a demand load miss which hit in the LLC.", + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to a demand load miss which hit in the LLC. If the cor= e has access to an L3 cache, an LLC hit refers to an L3 cache hit, otherwis= e it counts zeros.", "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0x34", "EventName": "MEM_BOUND_STALLS_LOAD.LLC_HIT", @@ -76,13 +163,21 @@ "UMask": "0x6" }, { - "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to a demand load miss which missed all the local cache= s.", + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to a demand load miss which missed all the local cache= s. If the core has access to an L3 cache, an LLC miss refers to an L3 cache= miss, otherwise it is an L2 cache miss.", "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0x34", "EventName": "MEM_BOUND_STALLS_LOAD.LLC_MISS", "SampleAfterValue": "1000003", "UMask": "0x78" }, + { + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled to a store buffer full condition", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x34", + "EventName": "MEM_BOUND_STALLS_LOAD.SBFULL", + "SampleAfterValue": "1000003", + "UMask": "0x80" + }, { "BriefDescription": "Counts the number of load ops retired that mi= ss the L3 cache and hit in DRAM", "Counter": "0,1,2,3,4,5,6,7", @@ -335,6 +430,33 @@ "SampleAfterValue": "200003", "UMask": "0x42" }, + { + "BriefDescription": "Counts the number of memory uops retired that= missed in the second level TLB.", + "Counter": "0,1,2,3,4,5,6,7", + "Data_LA": "1", + "EventCode": "0xd0", + "EventName": "MEM_UOPS_RETIRED.STLB_MISS", + "SampleAfterValue": "200003", + "UMask": "0x13" + }, + { + "BriefDescription": "Counts the number of load uops retired that m= iss in the second Level TLB.", + "Counter": "0,1,2,3,4,5,6,7", + "Data_LA": "1", + "EventCode": "0xd0", + "EventName": "MEM_UOPS_RETIRED.STLB_MISS_LOADS", + "SampleAfterValue": "200003", + "UMask": "0x11" + }, + { + "BriefDescription": "Counts the number of store uops retired that = miss in the second level TLB.", + "Counter": "0,1,2,3,4,5,6,7", + "Data_LA": "1", + "EventCode": "0xd0", + "EventName": "MEM_UOPS_RETIRED.STLB_MISS_STORES", + "SampleAfterValue": "200003", + "UMask": "0x12" + }, { "BriefDescription": "Counts the number of stores uops retired sam= e as MEM_UOPS_RETIRED.ALL_STORES", "Counter": "0,1,2,3,4,5,6,7", @@ -344,6 +466,16 @@ "SampleAfterValue": "1000003", "UMask": "0x6" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache where a snoop was sent, the snoop hit, and modified data was= forwarded.", "Counter": "0,1,2,3,4,5,6,7", @@ -364,6 +496,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache where a snoop was sent, the snoop hit, and modified data was= forwarded.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/grandridge/counter.json b/tools= /perf/pmu-events/arch/x86/grandridge/counter.json index 9fd5d8ad6d3b..d9ac3aca5bd5 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/counter.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/counter.json @@ -37,6 +37,6 @@ { "Unit": "CHACMS", "CountersNumFixed": "0", - "CountersNumGeneric": 4 + "CountersNumGeneric": "4" } ] \ No newline at end of file diff --git a/tools/perf/pmu-events/arch/x86/grandridge/frontend.json b/tool= s/perf/pmu-events/arch/x86/grandridge/frontend.json index 7cdf611efb23..fef5cba533bb 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/frontend.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/frontend.json @@ -31,5 +31,13 @@ "EventName": "ICACHE.MISSES", "SampleAfterValue": "200003", "UMask": "0x2" + }, + { + "BriefDescription": "Counts the number of cycles that the micro-se= quencer is busy.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xe7", + "EventName": "MS_DECODED.MS_BUSY", + "SampleAfterValue": "1000003", + "UMask": "0x4" } ] diff --git a/tools/perf/pmu-events/arch/x86/grandridge/grr-metrics.json b/t= ools/perf/pmu-events/arch/x86/grandridge/grr-metrics.json index 2f9959c61718..3029022e4e94 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/grr-metrics.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/grr-metrics.json @@ -69,7 +69,7 @@ }, { "BriefDescription": "Percentage of time spent in the active CPU po= wer state C0", - "MetricExpr": "tma_info_system_cpu_utilization", + "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", "MetricName": "cpu_utilization", "ScaleUnit": "100%" }, @@ -213,525 +213,6 @@ "MetricName": "stores_retired_per_instr", "ScaleUnit": "1per_instr" }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to certain allocation restrictions", - "MetricExpr": "tma_core_bound", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_core_bound_group", - "MetricName": "tma_allocation_restriction", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend due to backend stalls", - "MetricExpr": "TOPDOWN_BE_BOUND.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", - "MetricName": "tma_backend_bound", - "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend because allocation is stalled due to a misp= redicted jump or a machine clear", - "MetricExpr": "TOPDOWN_BAD_SPECULATION.ALL_P / (6 * CPU_CLK_UNHALT= ED.CORE)", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", - "MetricName": "tma_bad_speculation", - "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to BACLEARS, which occurs when the Branch T= arget Buffer (BTB) prediction or lack thereof, was corrected by a later bra= nch predictor in the frontend", - "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_DETECT / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", - "MetricName": "tma_branch_detect", - "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to branch mispredicts", - "MetricExpr": "TOPDOWN_BAD_SPECULATION.MISPREDICT / (6 * CPU_CLK_U= NHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_bad_speculation_g= roup", - "MetricName": "tma_branch_mispredicts", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to BTCLEARS, which occurs when the Branch T= arget Buffer (BTB) predicts a taken branch", - "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_RESTEER / (6 * CPU_CLK_UNHA= LTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", - "MetricName": "tma_branch_resteer", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to the microcode sequencer (MS)", - "MetricExpr": "TOPDOWN_FE_BOUND.CISC / (6 * CPU_CLK_UNHALTED.CORE)= ", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", - "MetricName": "tma_cisc", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of cycles due to backend bo= und stalls that are bounded by core restrictions and not attributed to an o= utstanding load or stores, or resource limitation", - "MetricExpr": "TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS / (6 * CPU_CLK_= UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_backend_bound_gro= up", - "MetricName": "tma_core_bound", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to decode stalls", - "MetricExpr": "TOPDOWN_FE_BOUND.DECODE / (6 * CPU_CLK_UNHALTED.COR= E)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", - "MetricName": "tma_decode", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to a machine clear that does not require the = use of microcode, classified as a fast nuke, due to memory ordering, memory= disambiguation and memory renaming", - "MetricExpr": "TOPDOWN_BAD_SPECULATION.FASTNUKE / (6 * CPU_CLK_UNH= ALTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_machine_clears_gr= oup", - "MetricName": "tma_fast_nuke", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to frontend stalls", - "MetricExpr": "TOPDOWN_FE_BOUND.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", - "MetricName": "tma_frontend_bound", - "MetricgroupNoGroup": "TopdownL1", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to instruction cache misses", - "MetricExpr": "TOPDOWN_FE_BOUND.ICACHE / (6 * CPU_CLK_UNHALTED.COR= E)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", - "MetricName": "tma_icache_misses", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend bandwidth restrictions due to d= ecode, predecode, cisc, and other limitations", - "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH / (6 * CPU_CLK_= UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_frontend_bound_gr= oup", - "MetricName": "tma_ifetch_bandwidth", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend latency restrictions due to ica= che misses, itlb misses, branch detection, and resteer limitations", - "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_LATENCY / (6 * CPU_CLK_UN= HALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_frontend_bound_gr= oup", - "MetricName": "tma_ifetch_latency", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Instructions per Floating Point (FP) Operatio= n", - "MetricExpr": "INST_RETIRED.ANY / FP_FLOPS_RETIRED.ALL", - "MetricGroup": "Flops", - "MetricName": "tma_info_arith_inst_mix_ipflop" - }, - { - "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction", - "MetricExpr": "INST_RETIRED.ANY / (FP_INST_RETIRED.128B_DP + FP_IN= ST_RETIRED.128B_SP)", - "MetricGroup": "Flops", - "MetricName": "tma_info_arith_inst_mix_ipfparith_avx128" - }, - { - "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction", - "MetricExpr": "INST_RETIRED.ANY / FP_INST_RETIRED.64B_DP", - "MetricGroup": "Flops", - "MetricName": "tma_info_arith_inst_mix_ipfparith_scalar_dp" - }, - { - "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction", - "MetricExpr": "INST_RETIRED.ANY / FP_INST_RETIRED.32B_SP", - "MetricGroup": "Flops", - "MetricName": "tma_info_arith_inst_mix_ipfparith_scalar_sp" - }, - { - "BriefDescription": "Percentage of time that retirement is stalled= due to a first level data TLB miss", - "MetricExpr": "100 * (LD_HEAD.DTLB_MISS_AT_RET + LD_HEAD.PGWALK_AT= _RET) / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "Cycles", - "MetricName": "tma_info_bottleneck_dtlb_miss_bound_cycles", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of time that allocation and retire= ment is stalled by the Frontend Cluster due to an Ifetch Miss, either Icach= e or ITLB Miss", - "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.ALL / CPU_CLK_UNHALTE= D.CORE", - "MetricGroup": "Cycles;Ifetch", - "MetricName": "tma_info_bottleneck_ifetch_miss_bound_cycles", - "PublicDescription": "Percentage of time that allocation and retir= ement is stalled by the Frontend Cluster due to an Ifetch Miss, either Icac= he or ITLB Miss. See Info.Ifetch_Bound", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of time that retirement is stalled= due to an L1 miss", - "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.ALL / CPU_CLK_UNHALTED.= CORE", - "MetricGroup": "Cycles;Load_Store_Miss", - "MetricName": "tma_info_bottleneck_load_miss_bound_cycles", - "PublicDescription": "Percentage of time that retirement is stalle= d due to an L1 miss. See Info.Load_Miss_Bound", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of time that retirement is stalled= by the Memory Cluster due to a pipeline stall", - "MetricExpr": "100 * LD_HEAD.ANY_AT_RET / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "Cycles;Mem_Exec", - "MetricName": "tma_info_bottleneck_mem_exec_bound_cycles", - "PublicDescription": "Percentage of time that retirement is stalle= d by the Memory Cluster due to a pipeline stall. See Info.Mem_Exec_Bound", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", - "MetricName": "tma_info_br_inst_mix_ipbranch" - }, - { - "BriefDescription": "Instruction per (near) call (lower number mea= ns higher occurrence rate)", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", - "MetricName": "tma_info_br_inst_mix_ipcall" - }, - { - "BriefDescription": "Instructions per Far Branch ( Far Branches ap= ply upon transition from application to operating system, handling interrup= ts, exceptions) [lower number means higher occurrence rate]", - "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", - "MetricName": "tma_info_br_inst_mix_ipfarbranch" - }, - { - "BriefDescription": "Instructions per retired conditional Branch M= isprediction where the branch was not taken", - "MetricExpr": "INST_RETIRED.ANY / (BR_MISP_RETIRED.COND - BR_MISP_= RETIRED.COND_TAKEN)", - "MetricName": "tma_info_br_inst_mix_ipmisp_cond_ntaken" - }, - { - "BriefDescription": "Instructions per retired conditional Branch M= isprediction where the branch was taken", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", - "MetricName": "tma_info_br_inst_mix_ipmisp_cond_taken" - }, - { - "BriefDescription": "Instructions per retired indirect call or jum= p Branch Misprediction", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", - "MetricName": "tma_info_br_inst_mix_ipmisp_indirect" - }, - { - "BriefDescription": "Instructions per retired return Branch Mispre= diction", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RETURN", - "MetricName": "tma_info_br_inst_mix_ipmisp_ret" - }, - { - "BriefDescription": "Instructions per retired Branch Misprediction= ", - "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", - "MetricName": "tma_info_br_inst_mix_ipmispredict" - }, - { - "BriefDescription": "Ratio of all branches which mispredict", - "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.ALL_= BRANCHES", - "MetricName": "tma_info_br_mispredict_bound_branch_mispredict_rati= o" - }, - { - "BriefDescription": "Ratio between Mispredicted branches and unkno= wn branches", - "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / BACLEARS.ANY", - "MetricName": "tma_info_br_mispredict_bound_branch_mispredict_to_u= nknown_branch_ratio" - }, - { - "BriefDescription": "Percentage of time that allocation is stalled= due to load buffer full", - "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.LD_BUF / CPU_CLK_UNHALTED= .CORE", - "MetricName": "tma_info_buffer_stalls_load_buffer_stall_cycles", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of time that allocation is stalled= due to memory reservation stations full", - "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.RSV / CPU_CLK_UNHALTED.CO= RE", - "MetricName": "tma_info_buffer_stalls_mem_rsv_stall_cycles", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of time that allocation is stalled= due to store buffer full", - "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.ST_BUF / CPU_CLK_UNHALTED= .CORE", - "MetricName": "tma_info_buffer_stalls_store_buffer_stall_cycles", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Cycles Per Instruction", - "MetricExpr": "CPU_CLK_UNHALTED.CORE / INST_RETIRED.ANY", - "MetricName": "tma_info_core_cpi", - "ScaleUnit": "1per_instr" - }, - { - "BriefDescription": "Floating Point Operations Per Cycle", - "MetricExpr": "FP_FLOPS_RETIRED.ALL / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "Flops", - "MetricName": "tma_info_core_flopc" - }, - { - "BriefDescription": "Instructions Per Cycle", - "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.CORE", - "MetricName": "tma_info_core_ipc" - }, - { - "BriefDescription": "Uops Per Instruction", - "MetricExpr": "TOPDOWN_RETIRING.ALL_P / INST_RETIRED.ANY", - "MetricName": "tma_info_core_upi" - }, - { - "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss hits in the L2", - "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.L2_HIT / MEM_BOUND_ST= ALLS_IFETCH.ALL", - "MetricName": "tma_info_ifetch_miss_bound_ifetchmissbound_with_l2h= it", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss hits in the L3", - "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.LLC_HIT / MEM_BOUND_S= TALLS_IFETCH.ALL", - "MetricName": "tma_info_ifetch_miss_bound_ifetchmissbound_with_l3h= it", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that hit the L2", - "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.L2_HIT / MEM_BOUND_STAL= LS_LOAD.ALL", - "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_miss_bound_loadmissbound_with_l2hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that hit the L3", - "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.LLC_HIT / MEM_BOUND_STA= LLS_LOAD.ALL", - "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_miss_bound_loadmissbound_with_l3hit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of cycles that the oldest l= oad of the load buffer is stalled at retirement due to a pipeline block", - "MetricExpr": "100 * LD_HEAD.L1_BOUND_AT_RET / CPU_CLK_UNHALTED.CO= RE", - "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_store_bound_l1_bound" - }, - { - "BriefDescription": "Counts the number of cycles that the oldest l= oad of the load buffer is stalled at retirement", - "MetricExpr": "100 * (LD_HEAD.L1_BOUND_AT_RET + MEM_BOUND_STALLS_L= OAD.ALL) / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_store_bound_load_bound" - }, - { - "BriefDescription": "Counts the number of cycles the core is stall= ed due to store buffer full", - "MetricExpr": "100 * (MEM_SCHEDULER_BLOCK.ST_BUF / MEM_SCHEDULER_B= LOCK.ALL) * tma_mem_scheduler", - "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_store_bound_store_bound" - }, - { - "BriefDescription": "Counts the number of machine clears relative = to thousands of instructions retired, due to floating point assists", - "MetricExpr": "1e3 * MACHINE_CLEARS.FP_ASSIST / INST_RETIRED.ANY", - "MetricName": "tma_info_machine_clear_bound_machine_clears_fp_assi= st_pki" - }, - { - "BriefDescription": "Counts the number of machine clears relative = to thousands of instructions retired, due to page faults", - "MetricExpr": "1e3 * MACHINE_CLEARS.PAGE_FAULT / INST_RETIRED.ANY", - "MetricName": "tma_info_machine_clear_bound_machine_clears_page_fa= ult_pki" - }, - { - "BriefDescription": "Counts the number of machine clears relative = to thousands of instructions retired, due to self-modifying code", - "MetricExpr": "1e3 * MACHINE_CLEARS.SMC / INST_RETIRED.ANY", - "MetricName": "tma_info_machine_clear_bound_machine_clears_smc_pki" - }, - { - "BriefDescription": "Percentage of total non-speculative loads wit= h an address aliasing block", - "MetricExpr": "100 * LD_BLOCKS.ADDRESS_ALIAS / MEM_UOPS_RETIRED.AL= L_LOADS", - "MetricName": "tma_info_mem_exec_blocks_loads_with_adressaliasing", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of total non-speculative loads wit= h a store forward or unknown store address block", - "MetricExpr": "100 * LD_BLOCKS.DATA_UNKNOWN / MEM_UOPS_RETIRED.ALL= _LOADS", - "MetricName": "tma_info_mem_exec_blocks_loads_with_storefwdblk", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of Memory Execution Bound due to a= first level data cache miss", - "MetricExpr": "100 * LD_HEAD.L1_MISS_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_l1miss", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of Memory Execution Bound due to o= ther block cases, such as pipeline conflicts, fences, etc", - "MetricExpr": "100 * LD_HEAD.OTHER_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_otherpipeline= blks", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of Memory Execution Bound due to a= pagewalk", - "MetricExpr": "100 * LD_HEAD.PGWALK_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_pagewalk", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of Memory Execution Bound due to a= second level TLB miss", - "MetricExpr": "100 * LD_HEAD.DTLB_MISS_AT_RET / LD_HEAD.ANY_AT_RET= ", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_stlbhit", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Percentage of Memory Execution Bound due to a= store forward address match", - "MetricExpr": "100 * LD_HEAD.ST_ADDR_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_storefwding", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Instructions per Load", - "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_LOADS", - "MetricName": "tma_info_mem_mix_ipload" - }, - { - "BriefDescription": "Instructions per Store", - "MetricExpr": "INST_RETIRED.ANY / MEM_UOPS_RETIRED.ALL_STORES", - "MetricName": "tma_info_mem_mix_ipstore" - }, - { - "BriefDescription": "Percentage of total non-speculative loads tha= t perform one or more locks", - "MetricExpr": "100 * MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRE= D.ALL_LOADS", - "MetricName": "tma_info_mem_mix_load_locks_ratio" - }, - { - "BriefDescription": "Percentage of total non-speculative loads tha= t are splits", - "MetricExpr": "100 * MEM_UOPS_RETIRED.SPLIT_LOADS / MEM_UOPS_RETIR= ED.ALL_LOADS", - "MetricName": "tma_info_mem_mix_load_splits_ratio" - }, - { - "BriefDescription": "Ratio of mem load uops to all uops", - "MetricExpr": "1e3 * MEM_UOPS_RETIRED.ALL_LOADS / TOPDOWN_RETIRING= .ALL_P", - "MetricName": "tma_info_mem_mix_memload_ratio" - }, - { - "BriefDescription": "Percentage of time that the core is stalled d= ue to a TPAUSE or UMWAIT instruction", - "MetricExpr": "100 * SERIALIZATION.C01_MS_SCB / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricName": "tma_info_serialization_tpause_cycles", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Average CPU Utilization", - "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / TSC", - "MetricName": "tma_info_system_cpu_utilization" - }, - { - "BriefDescription": "Giga Floating Point Operations Per Second", - "MetricExpr": "FP_FLOPS_RETIRED.ALL / (duration_time * 1e9)", - "MetricGroup": "Flops", - "MetricName": "tma_info_system_gflops", - "PublicDescription": "Giga Floating Point Operations Per Second. A= ggregate across all supported options of: FP precisions, scalar and vector = instructions, vector-width" - }, - { - "BriefDescription": "Fraction of cycles spent in Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.CORE_P:k / CPU_CLK_UNHALTED.CORE", - "MetricName": "tma_info_system_kernel_utilization" - }, - { - "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", - "MetricExpr": "CPU_CLK_UNHALTED.CORE_P / CPU_CLK_UNHALTED.CORE", - "MetricName": "tma_info_system_mux" - }, - { - "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", - "MetricExpr": "CPU_CLK_UNHALTED.CORE / CPU_CLK_UNHALTED.REF_TSC", - "MetricName": "tma_info_system_turbo_utilization" - }, - { - "BriefDescription": "Percentage of all uops which are FPDiv uops", - "MetricExpr": "100 * UOPS_RETIRED.FPDIV / TOPDOWN_RETIRING.ALL_P", - "MetricName": "tma_info_uop_mix_fpdiv_uop_ratio" - }, - { - "BriefDescription": "Percentage of all uops which are IDiv uops", - "MetricExpr": "100 * UOPS_RETIRED.IDIV / TOPDOWN_RETIRING.ALL_P", - "MetricName": "tma_info_uop_mix_idiv_uop_ratio" - }, - { - "BriefDescription": "Percentage of all uops which are microcode op= s", - "MetricExpr": "100 * UOPS_RETIRED.MS / TOPDOWN_RETIRING.ALL_P", - "MetricName": "tma_info_uop_mix_microcode_uop_ratio" - }, - { - "BriefDescription": "Percentage of all uops which are x87 uops", - "MetricExpr": "100 * UOPS_RETIRED.X87 / TOPDOWN_RETIRING.ALL_P", - "MetricName": "tma_info_uop_mix_x87_uop_ratio" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to Instruction Table Lookaside Buffer (ITLB= ) misses", - "MetricExpr": "TOPDOWN_FE_BOUND.ITLB_MISS / (6 * CPU_CLK_UNHALTED.= CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", - "MetricName": "tma_itlb_misses", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend because allocation is stalled due to a mach= ine clear (nuke) of any kind including memory ordering and memory disambigu= ation", - "MetricExpr": "TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS / (6 * CPU_C= LK_UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_bad_speculation_g= roup", - "MetricName": "tma_machine_clears", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to memory reservation stalls in which a sched= uler is not able to accept uops", - "MetricExpr": "TOPDOWN_BE_BOUND.MEM_SCHEDULER / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", - "MetricName": "tma_mem_scheduler", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to IEC or FPC RAT stalls, which can be due to= FIQ or IEC reservation stalls in which the integer, floating point or SIMD= scheduler is not able to accept uops", - "MetricExpr": "TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER / (6 * CPU_CLK_U= NHALTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", - "MetricName": "tma_non_mem_scheduler", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to a machine clear that requires the use of m= icrocode (slow nuke)", - "MetricExpr": "TOPDOWN_BAD_SPECULATION.NUKE / (6 * CPU_CLK_UNHALTE= D.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_machine_clears_gr= oup", - "MetricName": "tma_nuke", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to other common frontend stalls not categor= ized", - "MetricExpr": "TOPDOWN_FE_BOUND.OTHER / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", - "MetricName": "tma_other_fb", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to wrong predecodes", - "MetricExpr": "TOPDOWN_FE_BOUND.PREDECODE / (6 * CPU_CLK_UNHALTED.= CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", - "MetricName": "tma_predecode", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to the physical register file unable to accep= t an entry (marble stalls)", - "MetricExpr": "TOPDOWN_BE_BOUND.REGISTER / (6 * CPU_CLK_UNHALTED.C= ORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", - "MetricName": "tma_register", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to the reorder buffer being full (ROB stalls)= ", - "MetricExpr": "TOPDOWN_BE_BOUND.REORDER_BUFFER / (6 * CPU_CLK_UNHA= LTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", - "MetricName": "tma_reorder_buffer", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of cycles the core is stall= ed due to a resource limitation", - "MetricExpr": "tma_backend_bound - tma_core_bound", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_backend_bound_gro= up", - "MetricName": "tma_resource_bound", - "MetricgroupNoGroup": "TopdownL2", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that result = in retirement slots", - "MetricExpr": "TOPDOWN_RETIRING.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", - "MetricName": "tma_retiring", - "MetricgroupNoGroup": "TopdownL1", - "ScaleUnit": "100%" - }, - { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to scoreboards from the instruction queue (IQ= ), jump execution unit (JEU), or microcode sequencer (MS)", - "MetricExpr": "TOPDOWN_BE_BOUND.SERIALIZATION / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", - "MetricName": "tma_serialization", - "ScaleUnit": "100%" - }, { "BriefDescription": "Uncore operating frequency in GHz", "MetricExpr": "UNC_CHA_CLOCKTICKS / (source_count(UNC_CHA_CLOCKTIC= KS) * #num_packages) / 1e9 / duration_time", diff --git a/tools/perf/pmu-events/arch/x86/grandridge/other.json b/tools/p= erf/pmu-events/arch/x86/grandridge/other.json index 28f9a4c3ea84..daa16030d493 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/other.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/other.json @@ -8,26 +8,6 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3,4,5,6,7", @@ -37,13 +17,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4" } ] diff --git a/tools/perf/pmu-events/arch/x86/grandridge/pipeline.json b/tool= s/perf/pmu-events/arch/x86/grandridge/pipeline.json index 40fa4f5ae261..a934b64f66d0 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/pipeline.json @@ -56,6 +56,14 @@ "SampleAfterValue": "200003", "UMask": "0xfb" }, + { + "BriefDescription": "Counts the number of near indirect JMP branch= instructions retired.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xc4", + "EventName": "BR_INST_RETIRED.INDIRECT_JMP", + "SampleAfterValue": "200003", + "UMask": "0xef" + }, { "BriefDescription": "This event is deprecated. Refer to new event = BR_INST_RETIRED.INDIRECT_CALL", "Counter": "0,1,2,3,4,5,6,7", @@ -81,6 +89,30 @@ "SampleAfterValue": "200003", "UMask": "0xf7" }, + { + "BriefDescription": "Counts the number of near taken branch instru= ctions retired.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xc4", + "EventName": "BR_INST_RETIRED.NEAR_TAKEN", + "SampleAfterValue": "200003", + "UMask": "0xc0" + }, + { + "BriefDescription": "Counts the number of near relative CALL branc= h instructions retired.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xc4", + "EventName": "BR_INST_RETIRED.REL_CALL", + "SampleAfterValue": "200003", + "UMask": "0xfd" + }, + { + "BriefDescription": "Counts the number of near relative JMP branch= instructions retired.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xc4", + "EventName": "BR_INST_RETIRED.REL_JMP", + "SampleAfterValue": "200003", + "UMask": "0xdf" + }, { "BriefDescription": "Counts the total number of mispredicted branc= h instructions retired for all branch types.", "Counter": "0,1,2,3,4,5,6,7", @@ -121,6 +153,14 @@ "SampleAfterValue": "200003", "UMask": "0xfb" }, + { + "BriefDescription": "Counts the number of mispredicted near indire= ct JMP branch instructions retired.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xc5", + "EventName": "BR_MISP_RETIRED.INDIRECT_JMP", + "SampleAfterValue": "200003", + "UMask": "0xef" + }, { "BriefDescription": "Counts the number of mispredicted near taken = branch instructions retired.", "Counter": "0,1,2,3,4,5,6,7", @@ -236,8 +276,9 @@ "UMask": "0x20" }, { - "BriefDescription": "Counts the number of machine clears that flus= h the pipeline and restart the machine with the use of microcode due to SMC= , MEMORY_ORDERING, FP_ASSISTS, PAGE_FAULT, DISAMBIGUATION, and FPC_VIRTUAL_= TRAP.", + "BriefDescription": "This event is deprecated.", "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", "EventCode": "0xc3", "EventName": "MACHINE_CLEARS.SLOW", "SampleAfterValue": "20003", @@ -259,6 +300,14 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4" + }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend because allocation is stalled due to a mispredict= ed jump or a machine clear. [This event is alias to TOPDOWN_BAD_SPECULATION= .ALL_P]", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/grandridge/uncore-cache.json b/= tools/perf/pmu-events/arch/x86/grandridge/uncore-cache.json index 6a80cf6cbd36..b89ab6e5cfb5 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/uncore-cache.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/uncore-cache.json @@ -8,6 +8,16 @@ "PortMask": "0x000", "Unit": "CHACMS" }, + { + "BriefDescription": "Counts the number of cycles FAST trigger is r= eceived from the global FAST distress wire.", + "Counter": "0,1,2,3", + "EventCode": "0x34", + "EventName": "UNC_CHACMS_RING_SRC_THRTL", + "Experimental": "1", + "PerPkg": "1", + "PortMask": "0x000", + "Unit": "CHACMS" + }, { "BriefDescription": "Number of CHA clock cycles while the event is= enabled", "Counter": "0,1,2,3", @@ -530,6 +540,26 @@ "UMask": "0x4", "Unit": "CHA" }, + { + "BriefDescription": "Ingress (from CMS) Allocations : IRQ : Counts= number of allocations per cycle into the specified Ingress queue.", + "Counter": "0,1,2,3", + "EventCode": "0x13", + "EventName": "UNC_CHA_RxC_INSERTS.IRQ", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "CHA" + }, + { + "BriefDescription": "Ingress (from CMS) Occupancy : IRQ : Counts n= umber of entries in the specified Ingress queue in each cycle.", + "Counter": "0", + "EventCode": "0x11", + "EventName": "UNC_CHA_RxC_OCCUPANCY.IRQ", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "CHA" + }, { "BriefDescription": "All TOR Inserts", "Counter": "0,1,2,3", @@ -603,7 +633,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Data read opt prefetch from local IA that mis= s the cache", + "BriefDescription": "Data read opt prefetch from local IA", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_DRD_OPT_PREF", @@ -764,7 +794,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Last level cache prefetch read for ownership = from local IA that miss the cache", + "BriefDescription": "Last level cache prefetch read for ownership = from local IA", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_LLCPREFRFO", @@ -859,7 +889,7 @@ "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_MISS_DRD_OPT_PREF_LOCAL", "PerPkg": "1", - "PublicDescription": "TOR Inserts : DRd_Opt_Prefs issued by iA Cor= es that missed the LLC", + "PublicDescription": "TOR Inserts : Data read opt prefetch from lo= cal iA that missed the LLC targeting local memory", "UMask": "0xc8a6fe01", "Unit": "CHA" }, @@ -934,7 +964,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Read for ownership from local IA that miss th= e cache", + "BriefDescription": "Read for ownership from local IA that miss th= e LLC targeting local memory", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_MISS_RFO_LOCAL", @@ -954,7 +984,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Read for ownership prefetch from local IA tha= t miss the cache", + "BriefDescription": "Read for ownership prefetch from local IA tha= t miss the LLC targeting local memory", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_MISS_RFO_PREF_LOCAL", @@ -1024,7 +1054,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Read for ownership from local IA that miss th= e cache", + "BriefDescription": "Read for ownership from local IA", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_RFO", @@ -1034,7 +1064,7 @@ "Unit": "CHA" }, { - "BriefDescription": "Read for ownership prefetch from local IA tha= t miss the cache", + "BriefDescription": "Read for ownership prefetch from local IA", "Counter": "0,1,2,3", "EventCode": "0x35", "EventName": "UNC_CHA_TOR_INSERTS.IA_RFO_PREF", @@ -1406,7 +1436,6 @@ "Counter": "0", "EventCode": "0x36", "EventName": "UNC_CHA_TOR_OCCUPANCY.IA_DRD_OPT", - "Experimental": "1", "PerPkg": "1", "PublicDescription": "TOR Occupancy : DRd_Opts issued by iA Cores", "UMask": "0xc827ff01", diff --git a/tools/perf/pmu-events/arch/x86/grandridge/uncore-memory.json b= /tools/perf/pmu-events/arch/x86/grandridge/uncore-memory.json index e75b3050ccd5..6a11e5505957 100644 --- a/tools/perf/pmu-events/arch/x86/grandridge/uncore-memory.json +++ b/tools/perf/pmu-events/arch/x86/grandridge/uncore-memory.json @@ -188,6 +188,256 @@ "PublicDescription": "DRAM Clockticks", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH0_RANK0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH0_RANK1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH0_RANK2", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH0_RANK3", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH1_RANK0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x10", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH1_RANK1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x20", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH1_RANK2", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x40", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", + "Counter": "0,1,2,3", + "EventCode": "0x47", + "EventName": "UNC_M_POWERDOWN_CYCLES.SCH1_RANK3", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x80", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles a given rank is in Power Down Mod= e and all pages are closed", + "Counter": "0,1,2,3", + "EventCode": "0x88", + "EventName": "UNC_M_POWER_CHANNEL_PPD_CYCLES", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "MR4 temp reading is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.MR4BLKEN", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "RAPL is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.RAPLBLK", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, { "BriefDescription": "DRAM Precharge commands. : Counts the number = of DRAM Precharge commands sent on this channel.", "Counter": "0,1,2,3", @@ -360,6 +610,94 @@ "PerPkg": "1", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, { "BriefDescription": "Write Pending Queue Allocations", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 1b592cf63940..ed7a1845d43d 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -12,7 +12,7 @@ GenuineIntel-6-9[6C],v1.05,elkhartlake,core GenuineIntel-6-CF,v1.11,emeraldrapids,core GenuineIntel-6-5[CF],v13,goldmont,core GenuineIntel-6-7A,v1.01,goldmontplus,core -GenuineIntel-6-B6,v1.05,grandridge,core +GenuineIntel-6-B6,v1.07,grandridge,core GenuineIntel-6-A[DE],v1.06,graniterapids,core GenuineIntel-6-(3C|45|46),v36,haswell,core GenuineIntel-6-3F,v29,haswellx,core --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oo1-f73.google.com (mail-oo1-f73.google.com [209.85.161.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 2B5DC4EB51 for ; Wed, 26 Mar 2025 00:07:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.161.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947632; cv=none; b=sm8IW2Q38VktR1IKGD1JMRNAA9A6f0OtXtDQcEeKZMQgdqeqCjRFUxxhrRgi3PR3M2N1P+DXjx/H7Ud0ltgU2TMYCXWcLx0EiJve7ceZJjMHsUTIAsnb/0AVBYpf/6OTFf212+XVuw/h6J92mDNgYIjittj2E3jRs7OS1gK9ccQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947632; c=relaxed/simple; bh=VUw1fcLNkQyTdRlLD33gHLlCXcSYtZHb3F8Y5x/zUf0=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=POzwMK+v6nqS4GMhh3b0zlfo+j2OTjCw0VTeipiLdlUuARjuDeoGBKRXA50diPqjx2GTSq38Kt26eG0ZRjb2eg32uJ6dmuW/N6+uynPEHICsbCh+f00Rd2uW6cckY4h0C2FBUst7MIpyG1CwOmWVUozh8kCYy9KSJFAdCHmwwec= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=rUhmk+Kt; arc=none smtp.client-ip=209.85.161.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="rUhmk+Kt" Received: by mail-oo1-f73.google.com with SMTP id 006d021491bc7-6022020de0dso4079141eaf.0 for ; Tue, 25 Mar 2025 17:07:00 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947620; x=1743552420; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=Q2wBK9I1dTSU9BJYYyCMSSg8VMlpEyof7o+h6tqNxVI=; b=rUhmk+KtZQcU5R70DB600YHtl7FtuZXrg1myrDxU+yUuMA7w4miV8XGoCfbtVSuabB v0N97zhKxrKQzbLlKDaiYlA7SfTuV99xmqB0owjxLb0tHLjX4gfsOjBm2SjDgFdL2Zj5 pXz8KXJXQxbPjvOGM3uX4e1xVxem2gGjb1rI3mI+Qwx7WUk5TZFX69l9W25C3ObuZ4Lc K+MWMN1uB84mw7BlcHWUvoE7MuRONQVa/JrsjiAYaexTo9GDgeCrdxFVfOnMfjjU2s0W dtg1PQVXD9RpEEtGp8RLeAZyXZM79rAB+5BEh+oFEmOsMdeVjprM7K28RZ2hWCBiMHUX dZHA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947620; x=1743552420; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=Q2wBK9I1dTSU9BJYYyCMSSg8VMlpEyof7o+h6tqNxVI=; b=hz4chWLJFRYnozKQXeAZX/3lKPQdjGtXh7VPwMZZCtPzZoHoU8+eII8OnN8+FUuCV0 IRerGOgL46a6GxmKJEYSB0+KJM+XsURgdqHs7xtb75IylJizo2RF0HOSgciaux9tyQmU xdzIWLwZ+niVsa93oADcPhaDwPUKHde3RlL+dvneJ4evM4Qu1Ai1OmSMuVmRUzUCANkZ 3UlqJsik9p/afZFWfzF8VRV8bmXMMoEcRtxIEYYMV+DZ8dTARvpvDYEwMaasn0H7aJG7 BTyC4FnJrhJ+vwkN0S6wygcu8jMmPQSiAt3+787el0yGhj4m5LREYBjrFpdBLTpP7Y57 e86w== X-Forwarded-Encrypted: i=1; AJvYcCVHkYSRzRkA9sL1bZQ5sLNcZqOuS6KjzBKRVkFkFlnvMCY03VAQ0CCxr6xUggp3VCxJVeWleJsiUfBSvlE=@vger.kernel.org X-Gm-Message-State: AOJu0YyHY74Xk5q+OQTQm6Cx+UBNcx4zvfLfzrQ7/lWUFrmCgMHo/Z9P gMvDUwV6X/aHmWC3r1LJTeAoEQjYpkrfKpLRZRScMSNyUukWc5IJJ75ZKOQJJehR8jcn6r9yCeE sMO8qmQ== X-Google-Smtp-Source: AGHT+IHwbQwAFskD3C7Lx5ltZUdgytde1bfDXcGKcZyv0UMD/S71xpDtK6n9ZStpP8t+vXmr4reQwY3OEGCO X-Received: from oacpc10.prod.google.com ([2002:a05:6871:7a0a:b0:2c1:fc56:4d62]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6871:3147:b0:29f:e018:df85 with SMTP id 586e51a60fabf-2c780304fc0mr11359035fac.22.1742947620145; Tue, 25 Mar 2025 17:07:00 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:11 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-14-irogers@google.com> Subject: [PATCH v3 13/35] perf vendor events: Add graniterapids retirement latencies From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Add retirement latencies for use in place of retirement latency events. Update events from v1.06 to v1.08. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/graniterapids/cache.json | 122 +++++ .../arch/x86/graniterapids/counter.json | 5 + .../arch/x86/graniterapids/frontend.json | 21 + .../arch/x86/graniterapids/gnr-metrics.json | 483 +++++++++--------- .../arch/x86/graniterapids/memory.json | 130 +++++ .../arch/x86/graniterapids/other.json | 237 --------- .../arch/x86/graniterapids/pipeline.json | 52 ++ .../arch/x86/graniterapids/uncore-cache.json | 42 ++ .../graniterapids/uncore-interconnect.json | 90 +++- .../arch/x86/graniterapids/uncore-memory.json | 240 +++++++++ tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- 11 files changed, 926 insertions(+), 498 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/cache.json b/tool= s/perf/pmu-events/arch/x86/graniterapids/cache.json index d155da8610d8..6fa985656312 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/cache.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/cache.json @@ -351,6 +351,9 @@ "EventCode": "0xd0", "EventName": "MEM_INST_RETIRED.LOCK_LOADS", "PublicDescription": "Counts retired load instructions with locked= access.", + "RetirementLatencyMax": 5156, + "RetirementLatencyMean": 63.76, + "RetirementLatencyMin": 15, "SampleAfterValue": "100007", "UMask": "0x21" }, @@ -361,6 +364,9 @@ "EventCode": "0xd0", "EventName": "MEM_INST_RETIRED.SPLIT_LOADS", "PublicDescription": "Counts retired load instructions that split = across a cacheline boundary.", + "RetirementLatencyMax": 4704, + "RetirementLatencyMean": 3.97, + "RetirementLatencyMin": 0, "SampleAfterValue": "100003", "UMask": "0x41" }, @@ -371,6 +377,9 @@ "EventCode": "0xd0", "EventName": "MEM_INST_RETIRED.SPLIT_STORES", "PublicDescription": "Counts retired store instructions that split= across a cacheline boundary.", + "RetirementLatencyMax": 65535, + "RetirementLatencyMean": 19.0, + "RetirementLatencyMin": 0, "SampleAfterValue": "100003", "UMask": "0x42" }, @@ -381,6 +390,9 @@ "EventCode": "0xd0", "EventName": "MEM_INST_RETIRED.STLB_HIT_LOADS", "PublicDescription": "Number of retired load instructions with a c= lean hit in the 2nd-level TLB (STLB).", + "RetirementLatencyMax": 3424, + "RetirementLatencyMean": 1.57, + "RetirementLatencyMin": 0, "SampleAfterValue": "100003", "UMask": "0x9" }, @@ -391,6 +403,9 @@ "EventCode": "0xd0", "EventName": "MEM_INST_RETIRED.STLB_HIT_STORES", "PublicDescription": "Number of retired store instructions that hi= t in the 2nd-level TLB (STLB).", + "RetirementLatencyMax": 65535, + "RetirementLatencyMean": 5.24, + "RetirementLatencyMin": 0, "SampleAfterValue": "100003", "UMask": "0xa" }, @@ -430,6 +445,9 @@ "EventCode": "0xd2", "EventName": "MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD", "PublicDescription": "Counts retired load instructions whose data = sources were HitM responses from shared L3.", + "RetirementLatencyMax": 4472, + "RetirementLatencyMean": 353.04, + "RetirementLatencyMin": 0, "SampleAfterValue": "20011", "UMask": "0x4" }, @@ -440,6 +458,9 @@ "EventCode": "0xd2", "EventName": "MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS", "PublicDescription": "Counts the retired load instructions whose d= ata sources were L3 hit and cross-core snoop missed in on-pkg core cache.", + "RetirementLatencyMax": 830, + "RetirementLatencyMean": 125.27, + "RetirementLatencyMin": 0, "SampleAfterValue": "20011", "UMask": "0x1" }, @@ -460,6 +481,9 @@ "EventCode": "0xd2", "EventName": "MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD", "PublicDescription": "Counts retired load instructions whose data = sources were L3 and cross-core snoop hits in on-pkg core cache.", + "RetirementLatencyMax": 3939, + "RetirementLatencyMean": 289.9, + "RetirementLatencyMin": 0, "SampleAfterValue": "20011", "UMask": "0x2" }, @@ -470,6 +494,9 @@ "EventCode": "0xd3", "EventName": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM", "PublicDescription": "Retired load instructions which data sources= missed L3 but serviced from local DRAM.", + "RetirementLatencyMax": 4146, + "RetirementLatencyMean": 115.83, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x1" }, @@ -479,6 +506,9 @@ "Data_LA": "1", "EventCode": "0xd3", "EventName": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM", + "RetirementLatencyMax": 3572, + "RetirementLatencyMean": 430.22, + "RetirementLatencyMin": 0, "SampleAfterValue": "1000003", "UMask": "0x2" }, @@ -489,6 +519,9 @@ "EventCode": "0xd3", "EventName": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD", "PublicDescription": "Retired load instructions whose data sources= was forwarded from a remote cache.", + "RetirementLatencyMax": 8552, + "RetirementLatencyMean": 125.36, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x8" }, @@ -498,6 +531,9 @@ "Data_LA": "1", "EventCode": "0xd3", "EventName": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM", + "RetirementLatencyMax": 2580, + "RetirementLatencyMean": 135.29, + "RetirementLatencyMin": 0, "SampleAfterValue": "1000003", "UMask": "0x4" }, @@ -548,6 +584,9 @@ "EventCode": "0xd1", "EventName": "MEM_LOAD_RETIRED.L2_HIT", "PublicDescription": "Counts retired load instructions with L2 cac= he hits as data sources.", + "RetirementLatencyMax": 7140, + "RetirementLatencyMean": 5.71, + "RetirementLatencyMin": 0, "SampleAfterValue": "200003", "UMask": "0x2" }, @@ -568,6 +607,9 @@ "EventCode": "0xd1", "EventName": "MEM_LOAD_RETIRED.L3_HIT", "PublicDescription": "Counts retired load instructions with at lea= st one uop that hit in the L3 cache.", + "RetirementLatencyMax": 5630, + "RetirementLatencyMean": 57.64, + "RetirementLatencyMin": 0, "SampleAfterValue": "100021", "UMask": "0x4" }, @@ -598,6 +640,16 @@ "SampleAfterValue": "1000003", "UMask": "0x3" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit in the L3 or were snooped from another co= re's caches on the same socket.", "Counter": "0,1,2,3", @@ -618,6 +670,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit in the L3 o= r were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -678,6 +740,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that hit a modified = line in a distant L3 Cache or were snooped from a distant core's L1/L2 cach= es on this socket when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.SNC_CACHE.HITM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1008000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that either hit a no= n-modified line in a distant L3 Cache or were snooped from a distant core's= L1/L2 caches on this socket when the system is in SNC (sub-NUMA cluster) m= ode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.SNC_CACHE.HIT_WITH_FWD", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x808000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC0002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit in= the L3 or were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -698,6 +790,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10808", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC4477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit in the L3 or were snooped from another core's caches on the sa= me socket.", "Counter": "0,1,2,3", @@ -718,6 +830,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F33004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by a cache on a remote socket where a snoop was sent= and data was returned (Modified or Not Modified).", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/counter.json b/to= ols/perf/pmu-events/arch/x86/graniterapids/counter.json index 137da7efa8b1..5d3b202eadd3 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/counter.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/counter.json @@ -73,5 +73,10 @@ "Unit": "MDF", "CountersNumFixed": "0", "CountersNumGeneric": "4" + }, + { + "Unit": "UBOX", + "CountersNumFixed": "0", + "CountersNumGeneric": "2" } ] \ No newline at end of file diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/frontend.json b/t= ools/perf/pmu-events/arch/x86/graniterapids/frontend.json index dc81055941b1..77ebb46d104e 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/frontend.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/frontend.json @@ -53,6 +53,9 @@ "MSRIndex": "0x3F7", "MSRValue": "0x1", "PublicDescription": "Counts retired Instructions that experienced= DSB (Decode stream buffer i.e. the decoded instruction-cache) miss.", + "RetirementLatencyMax": 65535, + "RetirementLatencyMean": 2.46, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -75,6 +78,9 @@ "MSRIndex": "0x3F7", "MSRValue": "0x14", "PublicDescription": "Counts retired Instructions that experienced= iTLB (Instruction TLB) true miss.", + "RetirementLatencyMax": 980, + "RetirementLatencyMean": 41.96, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -86,6 +92,9 @@ "MSRIndex": "0x3F7", "MSRValue": "0x12", "PublicDescription": "Counts retired Instructions who experienced = Instruction L1 Cache true miss.", + "RetirementLatencyMax": 1785, + "RetirementLatencyMean": 9.83, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -97,6 +106,9 @@ "MSRIndex": "0x3F7", "MSRValue": "0x13", "PublicDescription": "Counts retired Instructions who experienced = Instruction L2 Cache true miss.", + "RetirementLatencyMax": 2854, + "RetirementLatencyMean": 137.41, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -250,6 +262,9 @@ "EventName": "FRONTEND_RETIRED.MS_FLOWS", "MSRIndex": "0x3F7", "MSRValue": "0x8", + "RetirementLatencyMax": 65535, + "RetirementLatencyMean": 77.14, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -261,6 +276,9 @@ "MSRIndex": "0x3F7", "MSRValue": "0x15", "PublicDescription": "Counts retired Instructions that experienced= STLB (2nd level TLB) true miss.", + "RetirementLatencyMax": 754, + "RetirementLatencyMean": 206.85, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, @@ -271,6 +289,9 @@ "EventName": "FRONTEND_RETIRED.UNKNOWN_BRANCH", "MSRIndex": "0x3F7", "MSRValue": "0x17", + "RetirementLatencyMax": 532, + "RetirementLatencyMean": 3.85, + "RetirementLatencyMin": 0, "SampleAfterValue": "100007", "UMask": "0x3" }, diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/gnr-metrics.json = b/tools/perf/pmu-events/arch/x86/graniterapids/gnr-metrics.json index a345b6874606..9d64d7f5f222 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/gnr-metrics.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/gnr-metrics.json @@ -310,7 +310,7 @@ "ScaleUnit": "1per_instr" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6) / (5 * tma_info_core_co= re_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -322,7 +322,7 @@ "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma= _core_bound_group", "MetricName": "tma_amx_busy", - "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tm= a_backend_bound > 0.2", + "MetricThreshold": "tma_amx_busy > 0.5 & (tma_core_bound > 0.1 & t= ma_backend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -330,12 +330,12 @@ "MetricExpr": "78 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -345,7 +345,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_inf= o_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -361,12 +361,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -381,7 +381,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -389,7 +389,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_= bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_d= ram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_st= ore_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_load= s + tma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_= l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_s= tore_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sha= ring + tma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_mem= ory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound = + tma_dram_bound + tma_store_bound)) * (tma_store_latency / (tma_store_late= ncy + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtl= b_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependen= cy + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_= bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma= _l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_fb= _full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tm= a_store_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tm= a_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_= latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_s= tore_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharin= g + tma_split_stores + tma_store_latency + tma_streaming_stores)) + tma_mem= ory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store= + tma_false_sharing + tma_split_stores + tma_store_latency + tma_streaming= _stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -397,22 +397,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_amx_busy + tma_ports_utilization) + tma_c= ore_bound * tma_amx_busy / (tma_divider + tma_serializing_operation + tma_a= mx_busy + tma_ports_utilization) + tma_core_bound * (tma_ports_utilization = / (tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utili= zation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_amx_busy= + tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_c= ore_bound * tma_amx_busy / (tma_amx_busy + tma_divider + tma_ports_utilizat= ion + tma_serializing_operation) + tma_core_bound * (tma_ports_utilization = / (tma_amx_busy + tma_divider + tma_ports_utilization + tma_serializing_ope= ration)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fe= tch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers= + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredict= s) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches= )) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_sw= itches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_= mite + tma_dsb + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetc= h_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers += tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts)= / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_branches))= / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_m= isses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms / (tma_ds= b + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_b= ranch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_othe= r_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_resteers + tma_c= lears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_itlb_mis= ses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) += tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ms)) + 10 * tma_m= icrocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_b= ranch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes = + tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_inf= o_thread_clks * tma_ports_utilized_0) / (tma_divider + tma_serializing_oper= ation + tma_amx_busy + tma_ports_utilization) + tma_microcode_sequencer / (= tma_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_m= icrocode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_switches + tma_bra= nch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_other_= mispredicts / tma_branch_mispredicts) / (tma_clears_resteers + tma_mispredi= cts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_dsb_swit= ches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) + t= ma_fetch_bandwidth * tma_ms / (tma_dsb + tma_mite + tma_ms)) + 10 * tma_mic= rocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_bra= nch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes + = tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_= thread_clks * tma_ports_utilized_0) / (tma_amx_busy + tma_divider + tma_por= ts_utilization + tma_serializing_operation) + tma_microcode_sequencer / (tm= a_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_mic= rocode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -420,7 +420,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_depende= ncy + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound= * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dra= m_bound + tma_store_bound)) * (tma_dtlb_store / (tma_store_latency + tma_fa= lse_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + = tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound= * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l= 3_bound + tma_store_bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false= _sharing + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -428,7 +428,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency + tma_streaming_stores - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -436,7 +436,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -448,10 +448,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -460,7 +460,7 @@ { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tm= a_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;Default;TmaL2;TopdownL2= ;tma_L2_group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -473,24 +473,24 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -498,8 +498,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -507,24 +507,24 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, FRONTEND_RETIRED.L1I_MISS * FRONTEND_RETIRED= .L1I_MISS:R / tma_info_thread_clks - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "FRONTEND_RETIRED.L2_MISS * FRONTEND_RETIRED.L2_MISS= :R / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -532,7 +532,7 @@ "MetricExpr": "max(0, FRONTEND_RETIRED.ITLB_MISS * FRONTEND_RETIRE= D.ITLB_MISS:R / tma_info_thread_clks - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -540,48 +540,48 @@ "MetricExpr": "FRONTEND_RETIRED.STLB_MISS * FRONTEND_RETIRED.STLB_= MISS:R / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * IT= LB_MISSES.WALK_COMPLETED_2M_4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISS= ES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks * IT= LB_MISSES.WALK_COMPLETED_4K / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.= WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches.", "MetricExpr": "BR_MISP_RETIRED.COND_NTAKEN_COST * BR_MISP_RETIRED.= COND_NTAKEN_COST:R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_nt_mispredicts", - "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by taken conditional branches", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by taken conditional branches.", "MetricExpr": "BR_MISP_RETIRED.COND_TAKEN_COST * BR_MISP_RETIRED.C= OND_TAKEN_COST:R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_mispredicts", - "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((min(MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * MEM_LOAD_= L3_HIT_RETIRED.XSNP_MISS:R, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * (79 * tma_i= nfo_system_core_frequency) - 4.4 * tma_info_system_core_frequency) if 0 < M= EM_LOAD_L3_HIT_RETIRED.XSNP_MISS:R else MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS *= (79 * tma_info_system_core_frequency) - 4.4 * tma_info_system_core_frequen= cy) + (min(MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * MEM_LOAD_L3_HIT_RETIRED.XSNP_= FWD:R, MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (81 * tma_info_system_core_freque= ncy) - 4.4 * tma_info_system_core_frequency) if 0 < MEM_LOAD_L3_HIT_RETIRED= .XSNP_FWD:R else MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (81 * tma_info_system_c= ore_frequency) - 4.4 * tma_info_system_core_frequency) * (OCR.DEMAND_DATA_R= D.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DA= TA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOA= D_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS * min(MEM_LOAD_L= 3_HIT_RETIRED.XSNP_MISS:R, 74.6 * tma_info_system_core_frequency) + MEM_LOA= D_L3_HIT_RETIRED.XSNP_FWD * min(MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R, 76.6 * = tma_info_system_core_frequency) * (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (= OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_= WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -592,24 +592,24 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "((min(MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * MEM_LOA= D_L3_HIT_RETIRED.XSNP_NO_FWD:R, MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * (79 *= tma_info_system_core_frequency) - 4.4 * tma_info_system_core_frequency) if= 0 < MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD:R else MEM_LOAD_L3_HIT_RETIRED.XSN= P_NO_FWD * (79 * tma_info_system_core_frequency) - 4.4 * tma_info_system_co= re_frequency) + (min(MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * MEM_LOAD_L3_HIT_RET= IRED.XSNP_FWD:R, MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (79 * tma_info_system_c= ore_frequency) - 4.4 * tma_info_system_core_frequency) if 0 < MEM_LOAD_L3_H= IT_RETIRED.XSNP_FWD:R else MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (79 * tma_inf= o_system_core_frequency) - 4.4 * tma_info_system_core_frequency) * (1 - OCR= .DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM += OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB= _HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD * min(MEM_LOAD= _L3_HIT_RETIRED.XSNP_NO_FWD:R, 74.6 * tma_info_system_core_frequency) + MEM= _LOAD_L3_HIT_RETIRED.XSNP_FWD * min(MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD:R, 74.= 6 * tma_info_system_core_frequency) * (1 - OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_= HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SN= OOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MI= SS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -618,7 +618,7 @@ "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%" }, @@ -627,7 +627,7 @@ "MetricExpr": "MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_cl= ks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -637,7 +637,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -645,34 +645,34 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(min(MEM_INST_RETIRED.STLB_HIT_LOADS * MEM_INST_RET= IRED.STLB_HIT_LOADS:R, MEM_INST_RETIRED.STLB_HIT_LOADS * 7) if 0 < MEM_INST= _RETIRED.STLB_HIT_LOADS:R else MEM_INST_RETIRED.STLB_HIT_LOADS * 7) / tma_i= nfo_thread_clks + tma_load_stlb_miss", + "MetricExpr": "MEM_INST_RETIRED.STLB_HIT_LOADS * min(MEM_INST_RETI= RED.STLB_HIT_LOADS:R, 7) / tma_info_thread_clks + tma_load_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(min(MEM_INST_RETIRED.STLB_HIT_STORES * MEM_INST_RE= TIRED.STLB_HIT_STORES:R, MEM_INST_RETIRED.STLB_HIT_STORES * 7) if 0 < MEM_I= NST_RETIRED.STLB_HIT_STORES:R else MEM_INST_RETIRED.STLB_HIT_STORES * 7) / = tma_info_thread_clks + tma_store_stlb_miss", + "MetricExpr": "MEM_INST_RETIRED.STLB_HIT_STORES * min(MEM_INST_RET= IRED.STLB_HIT_STORES:R, 7) / tma_info_thread_clks + tma_store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", - "MetricExpr": "(170 * tma_info_system_core_frequency * cpu@OCR.DEM= AND_RFO.L3_MISS\\,offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core= _frequency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", + "MetricExpr": "(170 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_MISS@offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core_frequ= ency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, @@ -693,7 +693,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -704,7 +704,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -722,7 +722,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -731,15 +731,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -747,8 +747,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED2.SCALAR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -756,8 +756,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIR= ED2.VECTOR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6= , tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -765,8 +765,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -774,8 +774,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -783,8 +783,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_= 2", "ScaleUnit": "100%" }, { @@ -795,27 +795,27 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%" }, { @@ -823,24 +823,24 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= ", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= .", "MetricExpr": "BR_MISP_RETIRED.INDIRECT_CALL_COST * BR_MISP_RETIRE= D.INDIRECT_CALL_COST:R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_call_mispredicts", - "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions.= ", "MetricExpr": "max((BR_MISP_RETIRED.INDIRECT_COST * BR_MISP_RETIRE= D.INDIRECT_COST:R - BR_MISP_RETIRED.INDIRECT_CALL_COST * BR_MISP_RETIRED.IN= DIRECT_CALL_COST:R) / tma_info_thread_clks, 0)", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_jump_mispredicts", - "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -851,28 +851,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -900,7 +900,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -908,7 +908,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -916,10 +916,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -980,11 +981,11 @@ "MetricExpr": "(FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.P= ORT_1 + FP_ARITH_DISPATCHED.PORT_5) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -997,8 +998,8 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, @@ -1011,7 +1012,7 @@ }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, @@ -1061,10 +1062,10 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node" + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node." }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired branches who got branch a= ddress clears", @@ -1073,7 +1074,7 @@ "MetricName": "tma_info_frontend_unknown_branches_ret" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -1091,7 +1092,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -1099,7 +1100,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -1107,7 +1108,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -1115,7 +1116,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -1123,7 +1124,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Pr= ecision instruction (lower number means higher occurrence rate)", @@ -1131,7 +1132,7 @@ "MetricGroup": "Flops;FpScalar;InsType;Server", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_hp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -1139,7 +1140,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1194,7 +1195,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 13", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1331,7 +1332,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1396,21 +1397,21 @@ "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system= _time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_dram_bw", - "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW" + "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW." }, { "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R= 2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_sys= tem_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", - "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW" + "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW." }, { "BriefDescription": "Average Off-core access BW for Reads-to-Core = (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_inf= o_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", - "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches" + "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches." }, { "BriefDescription": "STLB (2nd level TLB) code speculative misses = per kilo instruction (misses of any page-size that complete the page walk)", @@ -1452,8 +1453,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1474,18 +1475,18 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D0x1@", + "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" @@ -1548,14 +1549,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1566,14 +1566,14 @@ }, { "BriefDescription": "Average latency of data read request to exter= nal DRAM memory [in nanoseconds]", - "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / cha_0@event\\=3D0x0@", + "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / uncore_cha_0@event\\=3D0x1@", "MetricGroup": "MemOffcore;MemoryLat;Server;SoC", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data= -read prefetches" }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1599,7 +1599,7 @@ }, { "BriefDescription": "Socket actual clocks when any core is active = on that socket", - "MetricExpr": "cha_0@event\\=3D0x0@", + "MetricExpr": "uncore_cha_0@event\\=3D0x1@", "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, @@ -1629,7 +1629,7 @@ "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1638,15 +1638,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1656,13 +1655,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1678,14 +1677,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 9" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1694,7 +1693,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%" }, { @@ -1702,8 +1701,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1711,8 +1710,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 = + INT_VEC_RETIRED.VNNI_256) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1720,8 +1719,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1729,7 +1728,7 @@ "MetricExpr": "max((EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.= STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1738,7 +1737,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1747,16 +1746,16 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.= STALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles wit= h demand load accesses that hit the L2 cache under unloaded scenarios (poss= ibly L2 latency limited)", - "MetricExpr": "(min(MEM_LOAD_RETIRED.L2_HIT * MEM_LOAD_RETIRED.L2_= HIT:R, MEM_LOAD_RETIRED.L2_HIT * (4.4 * tma_info_system_core_frequency)) if= 0 < MEM_LOAD_RETIRED.L2_HIT:R else MEM_LOAD_RETIRED.L2_HIT * (4.4 * tma_in= fo_system_core_frequency)) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRE= D.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * min(MEM_LOAD_RETIRED.L2_H= IT:R, 4.4 * tma_info_system_core_frequency) * (1 + MEM_LOAD_RETIRED.FB_HIT = / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1765,17 +1764,17 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.S= TALLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(min(MEM_LOAD_RETIRED.L3_HIT * MEM_LOAD_RETIRED.L3_= HIT:R, MEM_LOAD_RETIRED.L3_HIT * (37 * tma_info_system_core_frequency) - 4.= 4 * tma_info_system_core_frequency) if 0 < MEM_LOAD_RETIRED.L3_HIT:R else M= EM_LOAD_RETIRED.L3_HIT * (37 * tma_info_system_core_frequency) - 4.4 * tma_= info_system_core_frequency) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIR= ED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "MEM_LOAD_RETIRED.L3_HIT * min(MEM_LOAD_RETIRED.L3_H= IT:R, 32.6 * tma_info_system_core_frequency) * (1 + MEM_LOAD_RETIRED.FB_HIT= / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1783,19 +1782,19 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "DefaultMetricgroupName": "TopdownL2", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1812,7 +1811,7 @@ "MetricExpr": "max(0, tma_dtlb_load - tma_load_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1820,31 +1819,31 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1852,7 +1851,7 @@ "MetricExpr": "MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * MEM_LOAD_L3_M= ISS_RETIRED.LOCAL_DRAM:R * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1861,7 +1860,7 @@ "MetricExpr": "MEM_INST_RETIRED.LOCK_LOADS * MEM_INST_RETIRED.LOCK= _LOADS:R / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1877,19 +1876,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling)", + "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling).", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma= _mem_bandwidth_group", "MetricName": "tma_mba_stalls", - "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2= & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mba_stalls > 0.1 & (tma_mem_bandwidth > 0.= 2 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0= .2)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1898,32 +1897,32 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Backend;Default;TmaL2;TopdownL2;tma_L2_group;tma_b= ackend_bound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / (tma_r= etiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1944,7 +1943,7 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * INT_= MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1958,17 +1957,17 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clk= s / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clks = / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1976,10 +1975,10 @@ }, { "BriefDescription": "This metric estimates the fraction of cycles = when the CPU was stalled due to switches of uop delivery to the Microcode S= equencer (MS)", - "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1\\,edge\\=3D= 0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", + "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D1\\,edge@ / (UO= PS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1989,7 +1988,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1997,7 +1996,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -2011,19 +2010,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -2032,7 +2031,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%" }, { @@ -2041,7 +2040,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_po= rts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_5, tma_po= rt_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2050,7 +2049,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2059,7 +2058,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= ts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= t_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2067,8 +2066,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL)) / tm= a_info_thread_clks if ARITH.DIV_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - EXE_= ACTIVITY.BOUND_ON_LOADS else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EX= E_ACTIVITY.2_3_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -2076,8 +2075,8 @@ "MetricExpr": "max(EXE_ACTIVITY.EXE_BOUND_0_PORTS - RESOURCE_STALL= S.SCOREBOARD, 0) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -2085,7 +2084,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -2095,8 +2094,8 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%" }, { @@ -2105,7 +2104,7 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, @@ -2114,8 +2113,8 @@ "MetricExpr": "(MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM * MEM_LOAD_L3= _MISS_RETIRED.REMOTE_HITM:R + MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD * MEM_LOA= D_L3_MISS_RETIRED.REMOTE_FWD:R) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_R= ETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { @@ -2123,22 +2122,22 @@ "MetricExpr": "MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM:R * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRE= D.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s.", "MetricExpr": "BR_MISP_RETIRED.RET_COST * BR_MISP_RETIRED.RET_COST= :R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ret_mispredicts", - "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispre= dicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & (tma_branch_mispr= edicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2151,7 +2150,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks += tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -2160,8 +2159,8 @@ "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%" }, { @@ -2170,26 +2169,26 @@ "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles hand= ling memory load split accesses - load that cross 64-byte cache line bounda= ry", - "MetricExpr": "(min(MEM_INST_RETIRED.SPLIT_LOADS * MEM_INST_RETIRE= D.SPLIT_LOADS:R, MEM_INST_RETIRED.SPLIT_LOADS * tma_info_memory_load_miss_r= eal_latency) if 0 < MEM_INST_RETIRED.SPLIT_LOADS:R else MEM_INST_RETIRED.SP= LIT_LOADS * tma_info_memory_load_miss_real_latency) / tma_info_thread_clks", + "MetricExpr": "MEM_INST_RETIRED.SPLIT_LOADS * min(MEM_INST_RETIRED= .SPLIT_LOADS:R, tma_info_memory_load_miss_real_latency) / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents rate of split store ac= cesses", - "MetricExpr": "(min(MEM_INST_RETIRED.SPLIT_STORES * MEM_INST_RETIR= ED.SPLIT_STORES:R, MEM_INST_RETIRED.SPLIT_STORES) if 0 < MEM_INST_RETIRED.S= PLIT_STORES:R else MEM_INST_RETIRED.SPLIT_STORES) / tma_info_thread_clks", + "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES * min(MEM_INST_RETIRE= D.SPLIT_STORES:R, 1) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -2197,7 +2196,7 @@ "MetricExpr": "(XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS) / tma_in= fo_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -2206,8 +2205,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -2215,8 +2214,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -2224,8 +2223,8 @@ "MetricExpr": "(MEM_STORE_RETIRED.L2_HIT * 10 * (1 - MEM_INST_RETI= RED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_= LOADS / MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE= _REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -2242,7 +2241,7 @@ "MetricExpr": "max(0, tma_dtlb_store - tma_store_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -2250,31 +2249,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2282,7 +2281,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -2291,7 +2290,7 @@ "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_cl= ks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%" }, @@ -2300,8 +2299,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/memory.json b/too= ls/perf/pmu-events/arch/x86/graniterapids/memory.json index 5da5a10275ba..deddfb4686e1 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/memory.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/memory.json @@ -194,6 +194,16 @@ "SampleAfterValue": "1000003", "UMask": "0x2" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the local socket's L1, L= 2, or L3 caches.", "Counter": "0,1,2,3", @@ -204,6 +214,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -214,6 +244,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were n= ot supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -224,6 +284,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -254,6 +334,56 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x70C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x733004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.WRITE_ESTIMATE.MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0xFBFF80822", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/other.json b/tool= s/perf/pmu-events/arch/x86/graniterapids/other.json index 8df37f303273..99fcdc341272 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/other.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/other.json @@ -16,204 +16,6 @@ "SampleAfterValue": "1000003", "UMask": "0x8" }, - { - "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb7", - "EventName": "EXE.AMX_BUSY", - "SampleAfterValue": "2000003", - "UMask": "0x2" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC0002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10808", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC4477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x70C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F33004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x733004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -224,45 +26,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.WRITE_ESTIMATE.MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0xFBFF80822", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7" - }, - { - "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/pipeline.json b/t= ools/perf/pmu-events/arch/x86/graniterapids/pipeline.json index da6478607984..8530d93849fa 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/pipeline.json @@ -154,6 +154,9 @@ "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0xc5", "EventName": "BR_MISP_RETIRED.COND_NTAKEN_COST", + "RetirementLatencyMax": 888, + "RetirementLatencyMean": 6.11, + "RetirementLatencyMin": 0, "SampleAfterValue": "400009", "UMask": "0x50" }, @@ -171,6 +174,9 @@ "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0xc5", "EventName": "BR_MISP_RETIRED.COND_TAKEN_COST", + "RetirementLatencyMax": 2750, + "RetirementLatencyMean": 5.09, + "RetirementLatencyMin": 0, "SampleAfterValue": "400009", "UMask": "0x41" }, @@ -197,6 +203,9 @@ "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0xc5", "EventName": "BR_MISP_RETIRED.INDIRECT_CALL_COST", + "RetirementLatencyMax": 703, + "RetirementLatencyMean": 15.56, + "RetirementLatencyMin": 0, "SampleAfterValue": "400009", "UMask": "0x42" }, @@ -205,6 +214,9 @@ "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0xc5", "EventName": "BR_MISP_RETIRED.INDIRECT_COST", + "RetirementLatencyMax": 1562, + "RetirementLatencyMean": 11.07, + "RetirementLatencyMin": 0, "SampleAfterValue": "100003", "UMask": "0xc0" }, @@ -239,6 +251,9 @@ "Counter": "0,1,2,3,4,5,6,7", "EventCode": "0xc5", "EventName": "BR_MISP_RETIRED.RET_COST", + "RetirementLatencyMax": 1082, + "RetirementLatencyMean": 32.37, + "RetirementLatencyMin": 9, "SampleAfterValue": "100007", "UMask": "0x48" }, @@ -401,6 +416,14 @@ "SampleAfterValue": "1000003", "UMask": "0x4" }, + { + "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb7", + "EventName": "EXE.AMX_BUSY", + "SampleAfterValue": "2000003", + "UMask": "0x2" + }, { "BriefDescription": "Cycles total of 1 uop is executed on all port= s and Reservation Station was not empty.", "Counter": "0,1,2,3,4,5,6,7", @@ -774,6 +797,35 @@ "SampleAfterValue": "100003", "UMask": "0x2" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7" + }, + { + "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1" + }, { "BriefDescription": "This event counts a subset of the Topdown Slo= ts event that were not consumed by the back-end pipeline due to lack of bac= k-end resources, as a result of memory subsystem delays, execution units li= mitations, or other conditions.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-cache.json= b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-cache.json index 53055986534d..b782f6d54fc2 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-cache.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-cache.json @@ -853,6 +853,16 @@ "UMask": "0x8", "Unit": "CHA" }, + { + "BriefDescription": "Ingress (from CMS) Allocations : IRQ : Counts= number of allocations per cycle into the specified Ingress queue.", + "Counter": "0,1,2,3", + "EventCode": "0x13", + "EventName": "UNC_CHA_RxC_INSERTS.IRQ", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "CHA" + }, { "BriefDescription": "Ingress (from CMS) Occupancy : IRQ : Counts n= umber of entries in the specified Ingress queue in each cycle.", "Counter": "0", @@ -863,6 +873,38 @@ "UMask": "0x1", "Unit": "CHA" }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking exclusive lines in the core's cache. Snoop filter capacity e= victions occur when the snoop filter is full and evicts an existing entry t= o track a new entry. Does not count clean evictions such as when a core's c= ache replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.E_STATE", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : E state", + "UMask": "0x2", + "Unit": "CHA" + }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking modified lines in the core's cache. Snoop filter capacity ev= ictions occur when the snoop filter is full and evicts an existing entry to= track a new entry. Does not count clean evictions such as when a core's ca= che replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.M_STATE", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : M state", + "UMask": "0x1", + "Unit": "CHA" + }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking shared lines in the core's cache. Snoop filter capacity evic= tions occur when the snoop filter is full and evicts an existing entry to t= rack a new entry. Does not count clean evictions such as when a core's cach= e replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.S_STATE", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : S state", + "UMask": "0x4", + "Unit": "CHA" + }, { "BriefDescription": "All TOR Inserts", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-interconne= ct.json b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-interconnect.= json index 5c50275c79b0..e5bd11b27bcd 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-interconnect.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-interconnect.json @@ -1076,7 +1076,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for AD_BNC", + "BriefDescription": "Egress bypasses for AD_BNC", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.AD_BNC", @@ -1086,7 +1086,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for AD_CRD", + "BriefDescription": "Egress bypasses for AD_CRD", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.AD_CRD", @@ -1096,7 +1096,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for AK", + "BriefDescription": "Egress bypasses for AK", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.AK", @@ -1106,7 +1106,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for BL_BNC", + "BriefDescription": "Egress bypasses for BL_BNC", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.BL_BNC", @@ -1116,7 +1116,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for BL_CRD", + "BriefDescription": "Egress bypasses for BL_CRD", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.BL_CRD", @@ -1126,7 +1126,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress bypasses for for IV", + "BriefDescription": "Egress bypasses for IV", "Counter": "0,1,2,3", "EventCode": "0x1E", "EventName": "UNC_MDF_TxR_BYPASS.IV", @@ -1136,7 +1136,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for AD_BNC", + "BriefDescription": "Number of egress inserts for AD_BNC", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.AD_BNC", @@ -1146,7 +1146,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for AD_CRD", + "BriefDescription": "Number of egress inserts for AD_CRD", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.AD_CRD", @@ -1156,7 +1156,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for AK", + "BriefDescription": "Number of egress inserts for AK", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.AK", @@ -1166,7 +1166,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for BL_BNC", + "BriefDescription": "Number of egress inserts for BL_BNC", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.BL_BNC", @@ -1176,7 +1176,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for BL_CRD", + "BriefDescription": "Number of egress inserts for BL_CRD", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.BL_CRD", @@ -1186,7 +1186,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Number of egress inserts for for IV", + "BriefDescription": "Number of egress inserts for IV", "Counter": "0,1,2,3", "EventCode": "0x1C", "EventName": "UNC_MDF_TxR_INSERTS.IV", @@ -1196,7 +1196,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for AD_BNC", + "BriefDescription": "Egress occupancy for AD_BNC", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.AD_BNC", @@ -1206,7 +1206,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for AD_CRD", + "BriefDescription": "Egress occupancy for AD_CRD", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.AD_CRD", @@ -1216,7 +1216,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for AK", + "BriefDescription": "Egress occupancy for AK", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.AK", @@ -1226,7 +1226,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for BL_BNC", + "BriefDescription": "Egress occupancy for BL_BNC", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.BL_BNC", @@ -1236,7 +1236,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for BL_CRD", + "BriefDescription": "Egress occupancy for BL_CRD", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.BL_CRD", @@ -1246,7 +1246,7 @@ "Unit": "MDF" }, { - "BriefDescription": "Egress occupancy for for IV", + "BriefDescription": "Egress occupancy for IV", "Counter": "0,1,2,3", "EventCode": "0x1D", "EventName": "UNC_MDF_TxR_OCCUPANCY.IV", @@ -1932,5 +1932,59 @@ "Experimental": "1", "PerPkg": "1", "Unit": "UPI" + }, + { + "BriefDescription": "Message Received : Doorbell", + "Counter": "0,1", + "EventCode": "0x42", + "EventName": "UNC_U_EVENT_MSG.DOORBELL_RCVD", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x8", + "Unit": "UBOX" + }, + { + "BriefDescription": "Message Received : Interrupt", + "Counter": "0,1", + "EventCode": "0x42", + "EventName": "UNC_U_EVENT_MSG.INT_PRIO", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Message Received : Interrupt : Interrupts", + "UMask": "0x10", + "Unit": "UBOX" + }, + { + "BriefDescription": "Message Received : IPI", + "Counter": "0,1", + "EventCode": "0x42", + "EventName": "UNC_U_EVENT_MSG.IPI_RCVD", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Message Received : IPI : Inter Processor Int= errupts", + "UMask": "0x4", + "Unit": "UBOX" + }, + { + "BriefDescription": "Message Received : MSI", + "Counter": "0,1", + "EventCode": "0x42", + "EventName": "UNC_U_EVENT_MSG.MSI_RCVD", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Message Received : MSI : Message Signaled In= terrupts - interrupts sent by devices (including PCIe via IOxAPIC) (Socket = Mode only)", + "UMask": "0x2", + "Unit": "UBOX" + }, + { + "BriefDescription": "Message Received : VLW", + "Counter": "0,1", + "EventCode": "0x42", + "EventName": "UNC_U_EVENT_MSG.VLW_RCVD", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Message Received : VLW : Virtual Logical Wir= e (legacy) message were received from Uncore.", + "UMask": "0x1", + "Unit": "UBOX" } ] diff --git a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-memory.jso= n b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-memory.json index 5f4783ff6ce5..b991f6e1afbe 100644 --- a/tools/perf/pmu-events/arch/x86/graniterapids/uncore-memory.json +++ b/tools/perf/pmu-events/arch/x86/graniterapids/uncore-memory.json @@ -188,6 +188,94 @@ "PublicDescription": "DRAM Clockticks", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, { "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", "Counter": "0,1,2,3", @@ -286,6 +374,70 @@ "PublicDescription": "-", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "MR4 temp reading is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.MR4BLKEN", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "RAPL is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.RAPLBLK", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, { "BriefDescription": "DRAM Precharge commands. : Counts the number = of DRAM Precharge commands sent on this channel.", "Counter": "0,1,2,3", @@ -478,6 +630,94 @@ "UMask": "0x1", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, { "BriefDescription": "Write Pending Queue Allocations", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index ed7a1845d43d..579b4fbd65d6 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -13,7 +13,7 @@ GenuineIntel-6-CF,v1.11,emeraldrapids,core GenuineIntel-6-5[CF],v13,goldmont,core GenuineIntel-6-7A,v1.01,goldmontplus,core GenuineIntel-6-B6,v1.07,grandridge,core -GenuineIntel-6-A[DE],v1.06,graniterapids,core +GenuineIntel-6-A[DE],v1.08,graniterapids,core GenuineIntel-6-(3C|45|46),v36,haswell,core GenuineIntel-6-3F,v29,haswellx,core GenuineIntel-6-7[DE],v1.24,icelake,core --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id F24C17E0E4 for ; Wed, 26 Mar 2025 00:07:09 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947635; cv=none; b=qVVKN6LigVU+8ZxvCQcLsQ88sg+sYhgPMkrUc6IGzOVNaB39EwS2uqGEIpV1mIdXoGOgcJc2cm8fsxurCEj09OBa0ozBFiH0bKvZA4b1wEnOcuurGc90j3eL7hjrM+6Sa7zFmtKzL5s7oo2zpm3GXy+segtep37qJxMsOEnxFYw= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947635; c=relaxed/simple; bh=KMrcEmWjay1vuPIKNWGmAQ7ellmM1MUzgyJMlgx3QDo=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=jDnh3v3ABJOtM5pOBGl/u6+2eOZMp1kgUBHDkiNN9A/QNAPtP/JB6mohYupSrmjImceZZW2FP03PDSPvnrMv6dAPGjQ44QH0YTK1xLhvKmrdfuV4cUwocaKFpIJfTa2we9h+FWLx/fvTF33LiC6uWp5qDfIFZ8vUoMUzv0mtUs8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=uItTX0Y3; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="uItTX0Y3" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff6af1e264so17530914a91.3 for ; Tue, 25 Mar 2025 17:07:09 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947629; x=1743552429; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=/xU0wBAsl4jlEkSKnJ2JM7pI7tPOhIu8nmWNVPVFxSY=; b=uItTX0Y3spBkZbBReBDQIHHg4pjRmzxHEWYKCIRCKAeDYWJI1QFpn5co7TKRoX6to7 rmxKd+snTjJ7cE+1NVbIz68rNPbRl2ykfjnjKD3VNjVDWVosdPY2GRxJjrinNOqBa4/3 tpQn7pV1k9tXIR0spzZXmxxJDqw8lgXfnukAdREMmW1aXTwYox4SqlYOecplzY5yqwO8 KGFcyxVTKVZUN41HlHKLI4m7mEKe6GPaAhbL+jIIP1Uda5A4Yh97U9XTljCujUBP+ltd bgNJ+A7oJnINZyGqNWJ/jvHPtVdoaXvvm374J5Pju8wg6/LAabZeuXg5+B9Pm7V9e2Ev wa+w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947629; x=1743552429; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=/xU0wBAsl4jlEkSKnJ2JM7pI7tPOhIu8nmWNVPVFxSY=; b=pPzCXVXmOnps/AjzloNXrfbrnVuHdMg98gOzAq9MozsHBlo2H47korSM1/Pw+xgfLf Enmi+lsLKPPBRW7f1EYs67N+1xZa0MYvfRtCqI1r7PEhZaHRCLaigkgDaCVLZMnYFM+X 1BodUfnoX2eMVptSHuPOSYyTjkiBqtH3KpCrHl4Qf//wakWzALONS628q4vvd7loQDjc fFB3Se5q8AKIch+hkt9ovEn3Hlxda+8dmQfJxmHkk+qa6TFkipPhy3pa66BEUz2E8Dfk cqZzik5IF52KhTv53iUEwB/oAWyaFqjPc+cmiMfPF3iS5G8mfHf4C12M6bT3HznaAmeZ JouQ== X-Forwarded-Encrypted: i=1; AJvYcCVULKQfPvXVXIDWSSyDyE0U5zU9UjGsf9iClzxhkj8jFydSx/HVkcs2yE6Wak+63QHy709KM+a3pSotV2c=@vger.kernel.org X-Gm-Message-State: AOJu0YxX9Q7+JluwC/ENizxjZ8yaaTkdWNCgk9kkrElJuArD1sWHWxoW lUJmVw53uHuOquq84evl4ubOkcPNtpcQUuML8C718Uv3/EDMvCBxnlRmjJ9T0TNVQkWIpk4jF28 jws4Qdg== X-Google-Smtp-Source: AGHT+IFIlEvaXi/OCfb0+ze0Qs/Zzr3OBSWB/XDawaBrfpOPmWoJDrYbBDL6vo5H0pYmRhr2cB0Hxh2pvWdb X-Received: from pjtd13.prod.google.com ([2002:a17:90b:4d:b0:301:b354:8d63]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:4a11:b0:2ff:7b28:a51c with SMTP id 98e67ed59e1d1-3030ff09e7fmr23230289a91.34.1742947629052; Tue, 25 Mar 2025 17:07:09 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:12 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-15-irogers@google.com> Subject: [PATCH v3 14/35] perf vendor events: Update haswell metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/haswell/hsw-metrics.json | 206 +++++++++--------- 1 file changed, 102 insertions(+), 104 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/haswell/hsw-metrics.json b/tool= s/perf/pmu-events/arch/x86/haswell/hsw-metrics.json index 0c1040b7e38c..b26ea70a3628 100644 --- a/tools/perf/pmu-events/arch/x86/haswell/hsw-metrics.json +++ b/tools/perf/pmu-events/arch/x86/haswell/hsw-metrics.json @@ -74,12 +74,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", @@ -92,8 +92,8 @@ "MetricExpr": "66 * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_= slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y_WB_ASSIST", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, { @@ -104,7 +104,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -114,7 +114,7 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { @@ -125,7 +125,7 @@ "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers", "ScaleUnit": "100%" }, { @@ -133,8 +133,8 @@ "MetricExpr": "12 * (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS= .COUNT + BACLEARS.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -143,8 +143,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -153,8 +153,8 @@ "MetricExpr": "(60 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * (1 = + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_= UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS= _L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LO= AD_UOPS_RETIRED.L3_MISS))) + 43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS *= (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_L= OAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + ME= M_LOAD_UOPS_RETIRED.L3_MISS)))) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MIS= S. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears= ", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { @@ -165,7 +165,7 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { @@ -174,8 +174,8 @@ "MetricExpr": "43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * (1 + = MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UO= PS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L= 3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD= _UOPS_RETIRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, t= ma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_contested_accesses, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -183,8 +183,8 @@ "MetricExpr": "10 * ARITH.DIVIDER_UOPS / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -193,8 +193,8 @@ "MetricExpr": "(1 - MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_= RETIRED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS)) * CYCLE_ACTIVITY.STALL= S_L2_PENDING / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS_PS", "ScaleUnit": "100%" }, { @@ -203,7 +203,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -211,7 +211,7 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Related metrics: tma_fetch_bandw= idth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, @@ -220,8 +220,8 @@ "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + DTLB_LOAD_MISSES.W= ALK_DURATION) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_dtlb_store", "ScaleUnit": "100%" }, { @@ -229,8 +229,8 @@ "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + DTLB_STORE_MISSES= .WALK_DURATION) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -238,18 +238,18 @@ "MetricExpr": "60 * OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_= CORE / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3= _HIT_RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE.= Related metrics: tma_contested_accesses, tma_data_sharing, tma_machine_cle= ars", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.REQUEST_FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.REQUEST_FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -279,33 +279,33 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound.", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses.", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -316,7 +316,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -328,7 +328,7 @@ }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "(UOPS_EXECUTED.CORE / 2 / (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D0x1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@= ) if #SMT_on else UOPS_EXECUTED.CORE / (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D= 0x1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@))", + "MetricExpr": "(UOPS_EXECUTED.CORE / 2 / (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@) if= #SMT_on else UOPS_EXECUTED.CORE / (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ /= 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@))", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -353,7 +353,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -398,7 +398,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, = tma_lcp" }, { @@ -502,8 +502,8 @@ "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -537,14 +537,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -592,7 +591,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -601,8 +600,7 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -628,14 +626,14 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + ITLB_MISSES.WALK_DURAT= ION) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_M= ISSES.WALK_COMPLETED", "ScaleUnit": "100%" }, @@ -644,8 +642,8 @@ "MetricExpr": "max((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.ST= ALLS_LDM_PENDING) - CYCLE_ACTIVITY.STALLS_L1D_PENDING) / tma_info_thread_cl= ks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metri= cs: tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports= _utilized_1", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -653,8 +651,8 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L1D_PENDING - CYCLE_ACTIVITY= .STALLS_L2_PENDING) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT_PS", "ScaleUnit": "100%" }, { @@ -663,8 +661,8 @@ "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_RETIR= ED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS) * CYCLE_ACTIVITY.STALLS_L2_P= ENDING / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { @@ -673,8 +671,8 @@ "MetricExpr": "29 * (MEM_LOAD_UOPS_RETIRED.L3_HIT * (1 + MEM_LOAD_= UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRE= D.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RET= IRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_UOPS_RET= IRED.L3_MISS))) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: = tma_branch_resteers, tma_mem_latency, tma_store_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS. Related metric= s: tma_mem_latency", "ScaleUnit": "100%" }, { @@ -682,18 +680,18 @@ "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage,= tma_info_inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -712,8 +710,8 @@ "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, { @@ -724,15 +722,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, = tma_microcode_sequencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sh= aring, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_c= ache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x6@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", "ScaleUnit": "100%" }, @@ -741,19 +739,19 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS= _LDM_PENDING) + RESOURCE_STALLS.SB) / (min(CPU_CLK_UNHALTED.THREAD, CYCLE_A= CTIVITY.CYCLES_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - (cp= u@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc > 1.8 else cpu= @UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma= _fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_C= LK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.C= ORE\\,cmask\\=3D0x1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info= _thread_ipc > 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) - (RS_EVENT= S.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) * t= ma_backend_bound", + "MetricExpr": "((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALL= S_LDM_PENDING) + RESOURCE_STALLS.SB) / (min(CPU_CLK_UNHALTED.THREAD, CYCLE_= ACTIVITY.CYCLES_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - (cpu= @UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else cpu@UO= PS_EXECUTED.CORE\\,cmask\\=3D2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetc= h_latency > 0.1 else 0) + RESOURCE_STALLS.SB) if #SMT_on else min(CPU_CLK_U= NHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\= \,cmask\\=3D1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_= ipc > 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) - (RS_EVENTS.EMPTY_CY= CLES if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) * tma_backend= _bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { @@ -762,7 +760,7 @@ "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operatio= ns_group;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_ope= rations > 0.1", - "PublicDescription": "This metric represents fraction of slots the= CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The M= S is used for CISC instructions not supported by the default decoders (like= repeat move strings; or CPUID); or by microcode assists used to address so= me operation modes (like in Floating Point assists). These cases can often = be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_l1_bound, tma_ma= chine_clears, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The M= S is used for CISC instructions not supported by the default decoders (like= repeat move strings; or CPUID); or by microcode assists used to address so= me operation modes (like in Floating Point assists). These cases can often = be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_clears_resteers,= tma_l1_bound, tma_machine_clears, tma_ms_switches", "ScaleUnit": "100%" }, { @@ -771,7 +769,7 @@ "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", "ScaleUnit": "100%" }, { @@ -779,8 +777,8 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_l1_bound= , tma_machine_clears, tma_microcode_sequencer", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mix= ing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, { @@ -789,7 +787,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_por= t_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -798,7 +796,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_port_0, tma_port_5, tma_port_6, tma_p= orts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -834,7 +832,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_port_= 0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -843,7 +841,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port= _0, tma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -858,46 +856,46 @@ { "BriefDescription": "This metric estimates fraction of cycles the = CPU performance was potentially limited due to Core computation issues (non= divider-related)", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLE= S_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - (cpu@UOPS_EXECUT= ED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc > 1.8 else cpu@UOPS_EXECUTE= D.CORE\\,cmask\\=3D0x2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latenc= y > 0.1 else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_CLK_UNHALTED.T= HREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\\,cmask\\= =3D0x1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc >= 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) - (RS_EVENTS.EMPTY_CYCLE= S if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) - RESOURCE_STALL= S.SB - min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS_LDM_PENDING)) / t= ma_info_thread_clks", + "MetricExpr": "(min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES= _NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - (cpu@UOPS_EXECUTED.= CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1= else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_CLK_UNHALTED.THREAD, = CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ -= (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else c= pu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) - (RS_EVENTS.EMPTY_CYCLES if tma_fetc= h_latency > 0.1 else 0) + RESOURCE_STALLS.SB - RESOURCE_STALLS.SB - min(CPU= _CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS_LDM_PENDING)) / tma_info_thread= _clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed no uops on any execution port (Logical Processor cycles since ICL= , Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\=3D0x1\\,cmask\\=3D0= x1@ / 2 if #SMT_on else min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_= NO_EXECUTE) - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0)) /= tma_info_core_core_clks", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\,cmask\\=3D1@ / 2 if= #SMT_on else (min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUT= E) - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0)) / tma_info= _core_core_clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re the CPU executed total of 1 uop per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x2@) / 2 if #SMT_on else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) / tma_info_co= re_core_clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D2@) / 2 if #SMT_on else (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) / tma_info_core_core= _clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 2 uops per cycle on all execution ports (Logical Process= or cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x3@) / 2 if #SMT_on else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@) / tma_info_co= re_core_clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D3@) / 2 if #SMT_on else (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@) / tma_info_core_core= _clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_port_0, tma_port_1, tma_port_5, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ / 2 if #SM= T_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@) / tma_info_core_core_clk= s", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ / 2 if #SMT_= on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -917,7 +915,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -925,8 +923,8 @@ "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_c= ore_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -934,7 +932,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", "ScaleUnit": "100%" }, @@ -943,8 +941,8 @@ "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -952,8 +950,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -962,8 +960,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f201.google.com (mail-pl1-f201.google.com [209.85.214.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 071D41DFCB for ; Wed, 26 Mar 2025 00:07:18 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.201 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947644; cv=none; b=auJGqfmbZb5dMHx8UkhU9lB8eQB8L4xyK1pB84xU3zBF9G9pt4VZZ3vv3O7V70/kVf8ayflgYKGK4GPb9EX3FvdCg5IJoybDzdnhdBMYCWfRtqwlscxRrTbXmpGbgSr6Ry0SghWHHyjAo9KORXYNt/VHTFIQZAJWQVwrOUDGDGQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947644; c=relaxed/simple; bh=wTU5XM3LI5p+dSGarl/OC9mxlNqTvrUAVBHbQX7VLT4=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=hVsl+xwemyh/7teRyiYswc3KEGKWE8G5vB/VwrpOsqNmbx3LiL3uVPObdNe8C9L9nG1W8TPFnAloo8tTbwKVvbdr9L8MbUtKKLmP7KkFguoWt5ON4WMbS8OESKq8ZlnfHZ42eCjTinUT3KRP0W9mh7RfST4fdFg7OvZVDPSGyfg= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=ZE5+44ah; arc=none smtp.client-ip=209.85.214.201 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="ZE5+44ah" Received: by mail-pl1-f201.google.com with SMTP id d9443c01a7336-2242ade807fso174364635ad.2 for ; Tue, 25 Mar 2025 17:07:18 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947638; x=1743552438; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=oUFxYjBrTebQnEtyzKgARQbe01ALjiE4LybwOB+BCD4=; b=ZE5+44ah/84r9AOTp5Lv6hxKxYK6WtZ8h8yYFX18Dnoh3UFe0+A1crylkr80UTEPC2 mRoGqK6lbnYFRkbwG8luTL7X51ehcGBpMGHfDXgf6TUS6uwwfy76GW1+dWjp0tx8xHp3 y1BN/lvbaRksX6hYCRE+ciw5kfv7sxSGcBAL3jPUrgn3NJlmIFq4z1Xk19PE4EDes9ko y301AN7HU+4bp7rEFbd/IMai3eP2GWO4qRh6S4sg4VR2bzhqW2Tj3AkBK3nxuP21/+jX gj7VweDft/5WFtdYnoZjFbBt4AgFvL33HxXVrmszgMhlp1b6xAiiXAAp7pURe49nzEr5 qPgA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947638; x=1743552438; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=oUFxYjBrTebQnEtyzKgARQbe01ALjiE4LybwOB+BCD4=; b=Ox8e8ekzQbeViq0lh7e7hx9e3jxc6IA0Q+V58v533Wt2MrNvCOkxX/KnEq2x8yawtC JBSNzYtOrHYZOR9FwQk7koJXXVsslP1cY2Gzs1nOzUzpvLKBQIQazoC4wY37K9+lIjig OYgRNjEXK0qMQicd82s5SSTmQALOgF+yv1Q7uN+MGT4bNHLSzy/NddMVCD6rlTbcKwea jzYn5wffrkX3QI8lWxVtuHX09iELN02RRr2YiRMseT2LZhSAGB7Btjox5G2Q7LSTzQZB s+XZ0WdNKqc07/XXav/FQb95Oozk0cZqxDahSWLExhkfM+57obWiBcRBaPjs2nPyKBhX 1Gdw== X-Forwarded-Encrypted: i=1; AJvYcCVZ6yHUA1BHsnIHnwH0kfmXbOBpXo07En86iriKhDiaw7TjEHaFtQE+ftNPP+P61TD8mDhQVJWud/1rutU=@vger.kernel.org X-Gm-Message-State: AOJu0YzytX3zHyF3XR8huBtJhitbyE8I2LtP90pinLhF8Fo6V7uqfr1G uLQS7o2bwI7/1spmyqNSoVpU/qawCZQBfIs1Avyh3HnGitcIzMsI+LhmS4/J00SFgRL84ey2txC je0iD/g== X-Google-Smtp-Source: AGHT+IHmAGVCPMRGtk42hP7pAHoc4qfM3I+LzswhHSIozfbJu8WYYGA0FpQ9lifluO1s2l5y7jZD5TBXout+ X-Received: from pfbiv13.prod.google.com ([2002:a05:6a00:66cd:b0:736:3cd5:ba3f]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:b84:b0:736:5b85:a911 with SMTP id d2e1a72fcca58-7390598e30cmr36226237b3a.8.1742947638365; Tue, 25 Mar 2025 17:07:18 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:13 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-16-irogers@google.com> Subject: [PATCH v3 15/35] perf vendor events: Update haswellx metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/haswellx/hsx-metrics.json | 222 +++++++++--------- 1 file changed, 110 insertions(+), 112 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/haswellx/hsx-metrics.json b/too= ls/perf/pmu-events/arch/x86/haswellx/hsx-metrics.json index 1a05b74be575..8245a98ad4b9 100644 --- a/tools/perf/pmu-events/arch/x86/haswellx/hsx-metrics.json +++ b/tools/perf/pmu-events/arch/x86/haswellx/hsx-metrics.json @@ -276,12 +276,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", @@ -294,8 +294,8 @@ "MetricExpr": "66 * OTHER_ASSISTS.ANY_WB_ASSIST / tma_info_thread_= slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y_WB_ASSIST", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, { @@ -306,7 +306,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -316,7 +316,7 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { @@ -327,7 +327,7 @@ "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Branch Misprediction. These slots are either wasted= by uops fetched from an incorrectly speculated program path; or stalls whe= n the out-of-order part of the machine needs to recover its state from a sp= eculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics:= tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers", "ScaleUnit": "100%" }, { @@ -335,8 +335,8 @@ "MetricExpr": "12 * (BR_MISP_RETIRED.ALL_BRANCHES + MACHINE_CLEARS= .COUNT + BACLEARS.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -345,8 +345,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -355,8 +355,8 @@ "MetricExpr": "(60 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM * (1 = + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_= UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS= _L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LO= AD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_D= RAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RET= IRED.REMOTE_FWD))) + 43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS * (1 + ME= M_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS= _RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_= HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_U= OPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM = + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED= .REMOTE_FWD)))) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MIS= S. Related metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears= , tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { @@ -367,7 +367,7 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { @@ -376,8 +376,8 @@ "MetricExpr": "43 * (MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT * (1 + = MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UO= PS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L= 3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD= _UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRA= M + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIR= ED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_UOPS_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_contested_accesses, t= ma_false_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_contested_accesses, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -385,8 +385,8 @@ "MetricExpr": "10 * ARITH.DIVIDER_UOPS / tma_info_core_core_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -395,8 +395,8 @@ "MetricExpr": "(1 - MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_= RETIRED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS)) * CYCLE_ACTIVITY.STALL= S_L2_PENDING / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_RE= TIRED.L3_MISS_PS", "ScaleUnit": "100%" }, { @@ -405,7 +405,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -413,7 +413,7 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Related metrics: tma_fetch_bandw= idth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, @@ -422,8 +422,8 @@ "MetricExpr": "(8 * DTLB_LOAD_MISSES.STLB_HIT + DTLB_LOAD_MISSES.W= ALK_DURATION) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_UOPS_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_dtlb_store", "ScaleUnit": "100%" }, { @@ -431,8 +431,8 @@ "MetricExpr": "(8 * DTLB_STORE_MISSES.STLB_HIT + DTLB_STORE_MISSES= .WALK_DURATION) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES. Related metrics: tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_UOPS_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -440,18 +440,18 @@ "MetricExpr": "(200 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_= HITM + 60 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE) / tma_info= _thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_UOPS_L3= _HIT_RETIRED.XSNP_HITM, MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM, OFFCORE_= RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE, OFFCORE_RESPONSE.DEMAND_RFO.LL= C_MISS.REMOTE_HITM. Related metrics: tma_contested_accesses, tma_data_shari= ng, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.REQUEST_FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.REQUEST_FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -481,33 +481,33 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound.", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "tma_microcode_sequencer", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses.", "MetricExpr": "ICACHE.IFDATA_STALL / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -518,7 +518,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -530,7 +530,7 @@ }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "(UOPS_EXECUTED.CORE / 2 / (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D0x1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@= ) if #SMT_on else UOPS_EXECUTED.CORE / (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D= 0x1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@))", + "MetricExpr": "(UOPS_EXECUTED.CORE / 2 / (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D1@ / 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@) if= #SMT_on else UOPS_EXECUTED.CORE / (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ /= 2 if #SMT_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@))", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -555,7 +555,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -600,7 +600,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage, = tma_lcp" }, { @@ -704,8 +704,8 @@ "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -739,14 +739,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -757,14 +756,14 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\= =3D0x182@ / cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filter_opc\\=3D0x182@", + "MetricExpr": "UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\=3D0x18= 2@ / UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\=3D0x182\\,thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" }, { "BriefDescription": "Average latency of data read request to exter= nal memory (in nanoseconds)", - "MetricExpr": "1e9 * (cbox@UNC_C_TOR_OCCUPANCY.MISS_OPCODE\\,filte= r_opc\\=3D0x182@ / cbox@UNC_C_TOR_INSERTS.MISS_OPCODE\\,filter_opc\\=3D0x18= 2@) / (tma_info_system_socket_clks / tma_info_system_time)", + "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / tma_info_system_time)", "MetricGroup": "Mem;MemoryLat;SoC", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetche= s. ([RKL+]memory-controller only)" @@ -814,7 +813,7 @@ "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -823,8 +822,7 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -850,14 +848,14 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "(14 * ITLB_MISSES.STLB_HIT + ITLB_MISSES.WALK_DURAT= ION) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: ITLB_M= ISSES.WALK_COMPLETED", "ScaleUnit": "100%" }, @@ -866,8 +864,8 @@ "MetricExpr": "max((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.ST= ALLS_LDM_PENDING) - CYCLE_ACTIVITY.STALLS_L1D_PENDING) / tma_info_thread_cl= ks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT. Related metri= cs: tma_machine_clears, tma_microcode_sequencer, tma_ms_switches, tma_ports= _utilized_1", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -875,8 +873,8 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L1D_PENDING - CYCLE_ACTIVITY= .STALLS_L2_PENDING) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_UOPS_RETIRED.L2_HIT_PS", "ScaleUnit": "100%" }, { @@ -885,8 +883,8 @@ "MetricExpr": "MEM_LOAD_UOPS_RETIRED.L3_HIT / (MEM_LOAD_UOPS_RETIR= ED.L3_HIT + 7 * MEM_LOAD_UOPS_RETIRED.L3_MISS) * CYCLE_ACTIVITY.STALLS_L2_P= ENDING / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { @@ -895,8 +893,8 @@ "MetricExpr": "41 * (MEM_LOAD_UOPS_RETIRED.L3_HIT * (1 + MEM_LOAD_= UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD_UOPS_RETIRE= D.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UOPS_L3_HIT_RET= IRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_LOAD_UOPS_L3_= MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM + MEM_L= OAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE= _FWD))) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT. Related metrics: = tma_branch_resteers, tma_mem_latency, tma_store_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_UOPS_RETIRED.L3_HIT_PS. Related metric= s: tma_mem_latency", "ScaleUnit": "100%" }, { @@ -904,18 +902,18 @@ "MetricExpr": "ILD_STALL.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_frontend_dsb_coverage,= tma_info_inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -933,8 +931,8 @@ "MetricExpr": "200 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM * (= 1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOA= D_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_UO= PS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM_= LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE= _DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_R= ETIRED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.LOCAL_DRAM", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.LOCAL_DRAM_PS", "ScaleUnit": "100%" }, { @@ -943,8 +941,8 @@ "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, { @@ -955,15 +953,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_contested_accesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, = tma_microcode_sequencer, tma_ms_switches, tma_remote_cache", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_clears_resteers, tma_contested_accesses, tma_data_sharing, tma_false_sh= aring, tma_l1_bound, tma_microcode_sequencer, tma_ms_switches, tma_remote_c= ache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x6@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", "ScaleUnit": "100%" }, @@ -972,19 +970,19 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_l3_hit_latency", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS= _LDM_PENDING) + RESOURCE_STALLS.SB) / (min(CPU_CLK_UNHALTED.THREAD, CYCLE_A= CTIVITY.CYCLES_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - (cp= u@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc > 1.8 else cpu= @UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma= _fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_C= LK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.C= ORE\\,cmask\\=3D0x1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info= _thread_ipc > 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) - (RS_EVENT= S.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) * t= ma_backend_bound", + "MetricExpr": "((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALL= S_LDM_PENDING) + RESOURCE_STALLS.SB) / (min(CPU_CLK_UNHALTED.THREAD, CYCLE_= ACTIVITY.CYCLES_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - (cpu= @UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else cpu@UO= PS_EXECUTED.CORE\\,cmask\\=3D2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetc= h_latency > 0.1 else 0) + RESOURCE_STALLS.SB) if #SMT_on else min(CPU_CLK_U= NHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\= \,cmask\\=3D1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_= ipc > 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) - (RS_EVENTS.EMPTY_CY= CLES if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) * tma_backend= _bound", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { @@ -993,7 +991,7 @@ "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_heavy_operatio= ns_group;tma_issueMC;tma_issueMS", "MetricName": "tma_microcode_sequencer", "MetricThreshold": "tma_microcode_sequencer > 0.05 & tma_heavy_ope= rations > 0.1", - "PublicDescription": "This metric represents fraction of slots the= CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The M= S is used for CISC instructions not supported by the default decoders (like= repeat move strings; or CPUID); or by microcode assists used to address so= me operation modes (like in Floating Point assists). These cases can often = be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_l1_bound, tma_ma= chine_clears, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU was retiring uops fetched by the Microcode Sequencer (MS) unit. The M= S is used for CISC instructions not supported by the default decoders (like= repeat move strings; or CPUID); or by microcode assists used to address so= me operation modes (like in Floating Point assists). These cases can often = be avoided. Sample with: IDQ.MS_UOPS. Related metrics: tma_clears_resteers,= tma_l1_bound, tma_machine_clears, tma_ms_switches", "ScaleUnit": "100%" }, { @@ -1002,7 +1000,7 @@ "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to the MITE pipeline (the legacy dec= ode pipeline). This pipeline is used for code that was not pre-cached in th= e DSB or LSD. For example; inefficiencies due to asymmetric decoders; use o= f long immediate or LCP can manifest as MITE fetch bandwidth bottleneck.", "ScaleUnit": "100%" }, { @@ -1010,8 +1008,8 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_l1_bound= , tma_machine_clears, tma_microcode_sequencer", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_clears_r= esteers, tma_l1_bound, tma_machine_clears, tma_microcode_sequencer, tma_mix= ing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, { @@ -1020,7 +1018,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_por= t_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1029,7 +1027,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_port_0, tma_port_5, tma_port_6, tma_p= orts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1065,7 +1063,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_port_= 0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1074,7 +1072,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_port= _0, tma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1089,46 +1087,46 @@ { "BriefDescription": "This metric estimates fraction of cycles the = CPU performance was potentially limited due to Core computation issues (non= divider-related)", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLE= S_NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - (cpu@UOPS_EXECUT= ED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc > 1.8 else cpu@UOPS_EXECUTE= D.CORE\\,cmask\\=3D0x2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latenc= y > 0.1 else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_CLK_UNHALTED.T= HREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\\,cmask\\= =3D0x1@ - (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ if tma_info_thread_ipc >= 1.8 else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) - (RS_EVENTS.EMPTY_CYCLE= S if tma_fetch_latency > 0.1 else 0) + RESOURCE_STALLS.SB) - RESOURCE_STALL= S.SB - min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS_LDM_PENDING)) / t= ma_info_thread_clks", + "MetricExpr": "(min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES= _NO_EXECUTE) + (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - (cpu@UOPS_EXECUTED.= CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D2@)) / 2 - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1= else 0) + RESOURCE_STALLS.SB if #SMT_on else min(CPU_CLK_UNHALTED.THREAD, = CYCLE_ACTIVITY.CYCLES_NO_EXECUTE) + cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ -= (cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ if tma_info_thread_ipc > 1.8 else c= pu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) - (RS_EVENTS.EMPTY_CYCLES if tma_fetc= h_latency > 0.1 else 0) + RESOURCE_STALLS.SB - RESOURCE_STALLS.SB - min(CPU= _CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.STALLS_LDM_PENDING)) / tma_info_thread= _clks", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed no uops on any execution port (Logical Processor cycles since ICL= , Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\=3D0x1\\,cmask\\=3D0= x1@ / 2 if #SMT_on else min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_= NO_EXECUTE) - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0)) /= tma_info_core_core_clks", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,inv\\,cmask\\=3D1@ / 2 if= #SMT_on else (min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.CYCLES_NO_EXECUT= E) - (RS_EVENTS.EMPTY_CYCLES if tma_fetch_latency > 0.1 else 0)) / tma_info= _core_core_clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re the CPU executed total of 1 uop per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x1@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x2@) / 2 if #SMT_on else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D0x1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@) / tma_info_co= re_core_clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D1@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D2@) / 2 if #SMT_on else (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D1@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@) / tma_info_core_core= _clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 2 uops per cycle on all execution ports (Logical Process= or cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x2@ - cpu@UOP= S_EXECUTED.CORE\\,cmask\\=3D0x3@) / 2 if #SMT_on else cpu@UOPS_EXECUTED.COR= E\\,cmask\\=3D0x2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@) / tma_info_co= re_core_clks", + "MetricExpr": "((cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D2@ - cpu@UOPS_= EXECUTED.CORE\\,cmask\\=3D3@) / 2 if #SMT_on else (cpu@UOPS_EXECUTED.CORE\\= ,cmask\\=3D2@ - cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@) / tma_info_core_core= _clks)", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_port_0, tma_port_1, tma_port_5, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", - "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@ / 2 if #SM= T_on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D0x3@) / tma_info_core_core_clk= s", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", + "MetricExpr": "(cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@ / 2 if #SMT_= on else cpu@UOPS_EXECUTED.CORE\\,cmask\\=3D3@) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1137,8 +1135,8 @@ "MetricExpr": "(200 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM *= (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_L= OAD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= UOPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + ME= M_LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMO= TE_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS= _RETIRED.REMOTE_FWD))) + 180 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD * = (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LO= AD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM= _LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOT= E_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_= RETIRED.REMOTE_FWD)))) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM= , MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_contested_= accesses, tma_data_sharing, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_UOPS_L3_M= ISS_RETIRED.REMOTE_HITM_PS;MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_FWD_PS. Rel= ated metrics: tma_contested_accesses, tma_data_sharing, tma_false_sharing, = tma_machine_clears", "ScaleUnit": "100%" }, { @@ -1146,8 +1144,8 @@ "MetricExpr": "310 * (MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM * = (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LO= AD_UOPS_RETIRED.L3_HIT + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_L3_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_L3_HIT_RETIRED.XSNP_MISS + MEM= _LOAD_UOPS_L3_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOT= E_DRAM + MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_L3_MISS_= RETIRED.REMOTE_FWD))) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_UOPS_L3= _MISS_RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_UOPS_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { @@ -1167,7 +1165,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1175,8 +1173,8 @@ "MetricExpr": "2 * MEM_UOPS_RETIRED.SPLIT_STORES / tma_info_core_c= ore_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_UOPS_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1184,7 +1182,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", "ScaleUnit": "100%" }, @@ -1193,8 +1191,8 @@ "MetricExpr": "RESOURCE_STALLS.SB / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_UOPS_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1202,8 +1200,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1212,8 +1210,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 1AF6524B28 for ; Wed, 26 Mar 2025 00:07:26 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947655; cv=none; b=hcWFYW2ju/iTaAYsLAWjjKLBFK+EzHLRVGqLhLo7l/0a6J1dST087tWMSse+1A2JvZfS7O/LkE0Q4b1wFl10cGfoopxULpyVgUr+UwyGUJZX17qI/0mEwfbNJl8ktaYUVGRKEZ6kcVvACueNkmCfRe2M2zG0QWnU0Gnwtn+Nshg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947655; c=relaxed/simple; bh=v/ReQe5prlLbIsdt0PORCPuX7vbeMk0uz2PnbUyHtrw=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=gKVFew9UEXQC4fvjPhv1ZP5tb23Tg2eRKYkRoqf3hm9cM1ktWLTDL2j/cBPsnOfP/onS06ZeSll058ZDyLhn9xenHOTNDTIweovmxieAcVsLcJisXWLUuGIXxTr+dxS4ZX+RGbP/RR0YrAHkK40HKDc6APnfMKvXBu52uHUOCRk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=wJymg+iC; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="wJymg+iC" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2265a09dbfcso161937085ad.0 for ; Tue, 25 Mar 2025 17:07:26 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947646; x=1743552446; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=asqUmkMdWggbcu2w7sAqFbNHL77vzOg63lQ9FcTOxkI=; b=wJymg+iCYLyMLcaICwjXmWkBcOeo8ZOSomplycjlYvnT1rAaRfNhc9NFgL/rzK+UI+ Z89eWPhMUkiSp6ZIRgAKVNioWkjlNiWO9S8qNB7I2qR+vw0YvTaDI4koYigjzbSokeZS mP4QOgRqklzvRMZAG2BKpUxov69PkheRKznnIuUtrw2BZjIo+gVV2Na8fg7b+Z7w0NKi UlEYfhA519mUpX1K9yXFC2DrZCxshYNPM8j0DpIzMrXe/bTRy8T4FW5KPXY8uAPWrCye /3AZloU/4VDAUPZRqbdrncPiUQCiheVlsCnmJoBLp33SG5rb09tK9Z8Dn7QB0nySvYgj ilZA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947646; x=1743552446; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=asqUmkMdWggbcu2w7sAqFbNHL77vzOg63lQ9FcTOxkI=; b=LTT+kicBouXwU8UGR9MjBRBNXtovvINi/soDT9K3vinGRAzfB/efB/ojp7jl72gxQi JKux85NPbR2GpW3WIDzS1g+rpWNILY2WUSnh8Bezf5o8YKgj5t/8f/LCpSDkmHiv+opd 06+ZVpQO9tPrZ0Chw0vWz31xa+TU8H8t/WkmGJDNjJ7GByNmSZY2barxrlhYlqjJbij+ EA9xw1eifXJeStvuDKQINSeYYUMIu5Z+4/G+S4b4SXizW2COsxBlfDWYBNA3RL6UnpwZ we2CEHnLS2Sb/shUhyObnXxdeo8Urb4cAFhZ8nvZmsDgEcSRRfQm105vMrn37sONazRV RsOA== X-Forwarded-Encrypted: i=1; AJvYcCV/9H9opSkBKHJGmREYK4VSTvG7CfApG9Tfm0d/lQHML/MiEskmyfuuU14q3iXjBlM5Q25ayfHYOhBhX4Q=@vger.kernel.org X-Gm-Message-State: AOJu0Yyzqzvo1HQ/YFqfVtRRFMnLS4ID0tI0csGJJ2SQEfm91DoaeqDw bNKLbTAypULDByF/VGM8Xsi5cW1tkI5yZML828AqBWEzfC3QrJWKOUGp2QbN5gHHzYpt51XCS2y cQeggdA== X-Google-Smtp-Source: AGHT+IHq4wQm9oTe6ScWRv1ncNS20lLPwGYDW1Hrqqvk08ZMyFiIxRAziDJJLIvj78703ZLK0KZWzbJtD+qi X-Received: from pgmt1.prod.google.com ([2002:a63:2241:0:b0:af2:2287:e91c]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:7011:b0:1f5:8655:3287 with SMTP id adf61e73a8af0-1fe434562bcmr35061870637.40.1742947646339; Tue, 25 Mar 2025 17:07:26 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:14 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-17-irogers@google.com> Subject: [PATCH v3 16/35] perf vendor events: Update icelake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/icelake/cache.json | 60 +++ .../arch/x86/icelake/icl-metrics.json | 385 +++++++++--------- .../pmu-events/arch/x86/icelake/memory.json | 160 ++++++++ .../pmu-events/arch/x86/icelake/other.json | 220 ---------- 4 files changed, 412 insertions(+), 413 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/icelake/cache.json b/tools/perf= /pmu-events/arch/x86/icelake/cache.json index 015f70f157d1..e7bb2ca6f183 100644 --- a/tools/perf/pmu-events/arch/x86/icelake/cache.json +++ b/tools/perf/pmu-events/arch/x86/icelake/cache.json @@ -445,6 +445,16 @@ "SampleAfterValue": "50021", "UMask": "0x20" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit a cacheline in the L3 where a snoop was s= ent or not.", "Counter": "0,1,2,3", @@ -505,6 +515,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit a cacheline= in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", @@ -565,6 +585,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit a = cacheline in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", @@ -625,6 +655,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that hit a cacheline in the L3 where a= snoop was sent or not.", "Counter": "0,1,2,3", @@ -655,6 +695,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that hit a cacheline in the L3 where a snoop was sent or n= ot.", "Counter": "0,1,2,3", @@ -715,6 +765,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that hit a cacheline in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/icelake/icl-metrics.json b/tool= s/perf/pmu-events/arch/x86/icelake/icl-metrics.json index 63e28a03dc60..c5bfdb2f288b 100644 --- a/tools/perf/pmu-events/arch/x86/icelake/icl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/icelake/icl-metrics.json @@ -89,12 +89,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6) / (4 * tma_info_core_core_= clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -106,7 +106,7 @@ "MetricExpr": "34 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, @@ -129,12 +129,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -149,7 +149,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -157,7 +157,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store)) + tma_memory_bound * (tma_stor= e_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tm= a_store_bound)) * (tma_store_latency / (tma_store_latency + tma_false_shari= ng + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores)) + tma_memory_bound * (tma_stor= e_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tm= a_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_false_sharing = + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -165,22 +165,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_br= anches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma= _ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms /= (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_br= anches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tm= a_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms /= (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_mispredicts_resteers += tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_it= lb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_clears_resteers + tma_= mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_= dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -188,7 +188,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -196,15 +196,15 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears" + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -216,17 +216,17 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions.", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES= / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_branch_instructions", @@ -248,8 +248,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -257,8 +257,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -266,24 +266,24 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -291,7 +291,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -299,33 +299,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((32.5 * tma_info_system_core_frequency - 3.5 * tma= _info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM + (27 * tm= a_info_system_core_frequency - 3.5 * tma_info_system_core_frequency) * MEM_= LOAD_L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RE= TIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(29 * tma_info_system_core_frequency * MEM_LOAD_L3_= HIT_RETIRED.XSNP_HITM + 23.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L= 1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -335,25 +335,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(27 * tma_info_system_core_frequency - 3.5 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT * (1 + MEM_LOA= D_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "23.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _HIT_RETIRED.XSNP_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_= MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -362,7 +362,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -372,7 +372,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -382,7 +382,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -390,26 +390,26 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -417,8 +417,8 @@ "MetricExpr": "32.5 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, { @@ -437,7 +437,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -447,7 +447,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -465,7 +465,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -474,15 +474,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -490,7 +490,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -499,7 +499,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -508,7 +508,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -517,7 +517,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -526,7 +526,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -538,17 +538,17 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D0x1@) / IDQ.MITE_UOPS", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D1@) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -556,8 +556,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -569,28 +569,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -619,7 +619,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -628,7 +628,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -637,10 +637,11 @@ { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -701,11 +702,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -718,20 +719,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -773,7 +774,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -791,7 +792,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -799,7 +800,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -807,7 +808,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -815,7 +816,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -823,7 +824,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -831,7 +832,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -886,7 +887,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 11", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1005,7 +1006,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1067,8 +1068,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1095,12 +1096,12 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1141,14 +1142,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1175,7 +1175,7 @@ "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_= clks", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1183,7 +1183,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1191,7 +1191,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1219,7 +1219,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1228,15 +1228,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1246,13 +1245,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1268,14 +1267,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 7.5" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1283,8 +1282,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1292,7 +1291,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1301,7 +1300,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1311,7 +1310,7 @@ "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + L1D_PEND_MISS.FB_FULL_PERIODS)= * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1320,7 +1319,7 @@ "MetricExpr": "3.5 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1330,17 +1329,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(12.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETI= RED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "9 * tma_info_system_core_frequency * (MEM_LOAD_RETI= RED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) = / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1348,18 +1347,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1376,7 +1375,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1384,31 +1383,31 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1417,7 +1416,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1427,7 +1426,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%" }, { @@ -1437,15 +1436,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1454,7 +1453,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1465,11 +1464,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", @@ -1491,7 +1490,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1506,24 +1505,24 @@ }, { "BriefDescription": "This metric represents fraction of cycles whe= re (only) 4 uops were delivered by the MITE pipeline", - "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D0x4@ - cpu@IDQ.MITE_= UOPS\\,cmask\\=3D0x5@) / tma_info_thread_clks", + "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D4@ - cpu@IDQ.MITE_UO= PS\\,cmask\\=3D5@) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_gr= oup", "MetricName": "tma_mite_4wide", - "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_f= etch_bandwidth > 0.2", + "MetricThreshold": "tma_mite_4wide > 0.05 & (tma_mite > 0.1 & tma_= fetch_bandwidth > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D0x1@ / tma_info_core_co= re_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D1@ / tma_info_core_core= _clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1534,7 +1533,7 @@ "MetricExpr": "3 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1543,7 +1542,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -1558,19 +1557,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1614,8 +1613,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1623,8 +1622,8 @@ "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=3D0x80@ / t= ma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1632,7 +1631,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1641,7 +1640,7 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= port_0, tma_port_1, tma_port_5, tma_port_6", "ScaleUnit": "100%" }, @@ -1650,14 +1649,14 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -1670,7 +1669,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1679,7 +1678,7 @@ "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clk= s", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUS= E_INST", "ScaleUnit": "100%" }, @@ -1689,7 +1688,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1698,8 +1697,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1707,7 +1706,7 @@ "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1716,8 +1715,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1726,8 +1725,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1735,8 +1734,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 10 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1753,7 +1752,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -1761,31 +1760,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1793,7 +1792,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -1802,7 +1801,7 @@ "MetricExpr": "10 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1811,8 +1810,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/icelake/memory.json b/tools/per= f/pmu-events/arch/x86/icelake/memory.json index abaf3f4f9d63..1455aaac37b1 100644 --- a/tools/perf/pmu-events/arch/x86/icelake/memory.json +++ b/tools/perf/pmu-events/arch/x86/icelake/memory.json @@ -176,6 +176,16 @@ "SampleAfterValue": "50021", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -186,6 +196,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that was not supplie= d by the L3 cache.", "Counter": "0,1,2,3", @@ -196,6 +226,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that was no= t supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -206,6 +256,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that was not supplied by the L3 cache.= ", "Counter": "0,1,2,3", @@ -216,6 +286,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -226,6 +316,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -236,6 +346,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.OTHER.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184008000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -246,6 +376,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.OTHER.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184008000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.STREAMING_WR.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000800", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that was not supplied= by the L3 cache.", "Counter": "0,1,2,3", @@ -256,6 +406,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.STREAMING_WR.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000800", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/icelake/other.json b/tools/perf= /pmu-events/arch/x86/icelake/other.json index a96b2a989d3f..141cd30a30af 100644 --- a/tools/perf/pmu-events/arch/x86/icelake/other.json +++ b/tools/perf/pmu-events/arch/x86/icelake/other.json @@ -26,186 +26,6 @@ "SampleAfterValue": "200003", "UMask": "0x20" }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that have any type of response.", "Counter": "0,1,2,3", @@ -216,26 +36,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.OTHER.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184008000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.OTHER.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184008000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -245,25 +45,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.STREAMING_WR.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000800", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.STREAMING_WR.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000800", - "SampleAfterValue": "100003", - "UMask": "0x1" } ] --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oo1-f73.google.com (mail-oo1-f73.google.com [209.85.161.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 5942D8634E for ; Wed, 26 Mar 2025 00:07:36 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.161.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947666; cv=none; b=pSsEiFTrlJ5aZnGWwiiRc3zJCCouY0+aJa5av/Qyhga1aBztuB+LXSbZVrLckbdv2CP7X8Q9UxhuU+tWJZiop1EQIZ0MEVwoXbd6qS1ocFK9UWB2Ub70giUj2/ZacWirkpZcxFpie7I0ln7uVXBykIGoLJG8XN67pzwHApOAfYY= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947666; c=relaxed/simple; bh=asqU3Djvb2tHXwieDpCQ5yTIEnY4hYJNiGyxwIRPdLM=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=PTaziA6gtP1qbpeRcebJSwYwG5mp5SswOogBYQidc1dWeCEWuThM+VzdnDZV0do0PYi9rGvOzkXDoiudIiokdae6M4tScSp88OH8PueCGWxF+YxfW/hodPDU8cm0RFFfjgLS7ESs57E/lo1MsFIjMHboNExkpzL8SDxNb+DvEKk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=hE+I6FvE; arc=none smtp.client-ip=209.85.161.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="hE+I6FvE" Received: by mail-oo1-f73.google.com with SMTP id 006d021491bc7-60047981020so4742639eaf.1 for ; Tue, 25 Mar 2025 17:07:36 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947655; x=1743552455; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=2UpMjr96wUzX8tbHdnpAMHgCTlp+9vrgLW8XT5ILgPs=; b=hE+I6FvEf8lm7RSdZPCVyJxATOtDod+69X68tPB8mjlBu8THyO+I/ImWUm9YOIDnqQ QRF3vxLb0C4CCIXB15pbFtPpvZSv/yFVw2XnWQ7tSo2NUxa3wjhKJBpgk6DdH/tUTk5i /o0cSN/Ty99N1tXsrQMig8GI1HVUY55ZiIQl2BVvGc3/l3kUDIamhhF7kDQbxWU7z3m/ rGF+dWfjnpOatnLGw42CSjcRYYOxMbr2uEsPI7j+VgH+y5VpXELThU80IOS+f8Kw6lVC czs+Xudme+lIgqx/8jHQqNKgLwFrmOy9lpab0ohSENlI4NBHYfH0B/ir/BqKq4JdIvpk ZBbA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947655; x=1743552455; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=2UpMjr96wUzX8tbHdnpAMHgCTlp+9vrgLW8XT5ILgPs=; b=SEcpdP5P5GEL7eZa0AS+eWjBg+1yRknqPdrCdNqyGblCw+YClkgXlRWssY9ww+3YcD euVKIrBUvKZRBdUH6gjPIcqo15jFJRxDA/zMlP8+j6g5rDODYyIpZ3FFfnt70NSVIqu/ qg+jlXZrcxil8CYDqodPeDwgzvjxTkOwpMAHlkuxW0xwV3qEBzbmfqjphc+2klx5LRex kdTa+Ge+97kCmnGHON96KsQIKlH3Av0yur5MfoRcRumEX89FzI7ZKyWtkjcCwLXg3otU GzMJRlnFo27ilQRjqGL8sVHXlRmNNjcZGzoj4q7phwj7zLmFTB3SYNe9FcoPJjMKCaeA GznA== X-Forwarded-Encrypted: i=1; AJvYcCX6WBr6go9Uqf6xq9l+BiTb7qorO65vgr4m1OUcJNw+CUos+5ea6ZlP3m75tNykbgabY7RAkZKIaKj+mQc=@vger.kernel.org X-Gm-Message-State: AOJu0Yyhbqd/PAjzFq5iHWgYnAJSzXkJAWfXpe8JTmAGeOM3L3i+zQoH D03KXIR+qJCXCYARablC+B3QH+Vj6PizL4QtY/bywa5WnuRXINegiSKLzxQrVyQ/cxREZtZJICz 2lcT/dg== X-Google-Smtp-Source: AGHT+IGw/E+MAXH6+Fo5f36ciyGRBZIoUFtsRSoZPGBguAzfv/EUDlxQ/ezRcd2epAx3gpybjedoXwgNzIj8 X-Received: from oabmt9.prod.google.com ([2002:a05:6870:6b09:b0:2bc:69a2:7ea5]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6870:56a0:b0:2bc:61b3:b0ec with SMTP id 586e51a60fabf-2c780289995mr12493161fac.9.1742947655641; Tue, 25 Mar 2025 17:07:35 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:15 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-18-irogers@google.com> Subject: [PATCH v3 17/35] perf vendor events: Update icelakex events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/icelakex/cache.json | 273 +++++++++++ .../arch/x86/icelakex/icx-metrics.json | 399 ++++++++------- .../pmu-events/arch/x86/icelakex/memory.json | 190 +++++++ .../pmu-events/arch/x86/icelakex/other.json | 463 ------------------ 4 files changed, 662 insertions(+), 663 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/icelakex/cache.json b/tools/per= f/pmu-events/arch/x86/icelakex/cache.json index e8ab6ef2cd50..e46fd6f91d6b 100644 --- a/tools/perf/pmu-events/arch/x86/icelakex/cache.json +++ b/tools/perf/pmu-events/arch/x86/icelakex/cache.json @@ -1,4 +1,67 @@ [ + { + "BriefDescription": "Hit snoop reply with data, line invalidated.", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.I_FWD_FE", + "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated: removed from this core's cache, after the dat= a is forwarded back to the requestor and indicating the data was found unmo= dified in the (FE) Forward or Exclusive State in this cores caches cache. = A single snoop response from the core counts on all hyperthreads of the cor= e.", + "SampleAfterValue": "1000003", + "UMask": "0x20" + }, + { + "BriefDescription": "HitM snoop reply with data, line invalidated.= ", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.I_FWD_M", + "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated: removed from this core's caches, after the da= ta is forwarded back to the requestor, and indicating the data was found mo= dified(M) in this cores caches cache (aka HitM response). A single snoop r= esponse from the core counts on all hyperthreads of the core.", + "SampleAfterValue": "1000003", + "UMask": "0x10" + }, + { + "BriefDescription": "Hit snoop reply without sending the data, lin= e invalidated.", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.I_HIT_FSE", + "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated in this core's caches without being forwarded = back to the requestor. The line was in Forward, Shared or Exclusive (FSE) s= tate in this cores caches. A single snoop response from the core counts on= all hyperthreads of the core.", + "SampleAfterValue": "1000003", + "UMask": "0x2" + }, + { + "BriefDescription": "Line not found snoop reply", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.MISS", + "PublicDescription": "Counts responses to snoops indicating that t= he data was not found (IHitI) in this core's caches. A single snoop respons= e from the core counts on all hyperthreads of the Core.", + "SampleAfterValue": "1000003", + "UMask": "0x1" + }, + { + "BriefDescription": "Hit snoop reply with data, line kept in Share= d state.", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.S_FWD_FE", + "PublicDescription": "Counts responses to snoops indicating the li= ne may be kept on this core in the (S)hared state, after the data is forwar= ded back to the requestor, initially the data was found in the cache in the= (FS) Forward or Shared state. A single snoop response from the core count= s on all hyperthreads of the core.", + "SampleAfterValue": "1000003", + "UMask": "0x40" + }, + { + "BriefDescription": "HitM snoop reply with data, line kept in Shar= ed state", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.S_FWD_M", + "PublicDescription": "Counts responses to snoops indicating the li= ne may be kept on this core in the (S)hared state, after the data is forwar= ded back to the requestor, initially the data was found in the cache in the= (M)odified state. A single snoop response from the core counts on all hyp= erthreads of the core.", + "SampleAfterValue": "1000003", + "UMask": "0x8" + }, + { + "BriefDescription": "Hit snoop reply without sending the data, lin= e kept in Shared state.", + "Counter": "0,1,2,3", + "EventCode": "0xef", + "EventName": "CORE_SNOOP_RESPONSE.S_HIT_FSE", + "PublicDescription": "Counts responses to snoops indicating the li= ne was kept on this core in the (S)hared state, and that the data was found= unmodified but not forwarded back to the requestor, initially the data was= found in the cache in the (FSE) Forward, Shared state or Exclusive state. = A single snoop response from the core counts on all hyperthreads of the co= re.", + "SampleAfterValue": "1000003", + "UMask": "0x4" + }, { "BriefDescription": "Counts the number of cache lines replaced in = L1 data cache.", "Counter": "0,1,2,3", @@ -506,6 +569,16 @@ "SampleAfterValue": "100003", "UMask": "0x80" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit in the L3 or were snooped from another co= re's caches on the same socket.", "Counter": "0,1,2,3", @@ -546,6 +619,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit in the L3 o= r were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -586,6 +669,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In SN= C Mode counts only those PMM accesses that are controlled by the close SNC = Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703C00001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y a cache on a remote socket where a snoop hit a modified line in another c= ore's caches which forwarded the data.", "Counter": "0,1,2,3", @@ -606,6 +709,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit a modified = line in a distant L3 Cache or were snooped from a distant core's L1/L2 cach= es on this socket when the system is in SNC (sub-NUMA cluster) mode.", "Counter": "0,1,2,3", @@ -626,6 +739,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM on a distant memory controller of this socket when the system is in S= NC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SNC_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700800001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC0002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit in= the L3 or were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -646,6 +779,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM attached to this socket, unless in Sub NUMA Cluster(SNC) Mod= e. In SNC Mode counts only those PMM accesses that are controlled by the c= lose SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.LOCAL_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703C00002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.REMOTE_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit a = modified line in a distant L3 Cache or were snooped from a distant core's L= 1/L2 caches on this socket when the system is in SNC (sub-NUMA cluster) mod= e.", "Counter": "0,1,2,3", @@ -666,6 +829,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM on a distant memory controller of this socket when the syste= m is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SNC_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700800002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that hit in the L3 or were snooped fro= m another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -676,6 +849,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch (which bring data to= L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10070", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L3.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x12380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t hit in the L3 or were snooped from another core's caches on the same sock= et.", "Counter": "0,1,2,3", @@ -686,6 +879,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L3.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x90002380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts full cacheline writes (ItoM) that were= not supplied by the local socket's L1, L2, or L3 caches and the cacheline = was homed in a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.ITOM.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x90000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware and software prefetches to al= l cache levels that hit in the L3 or were snooped from another core's cache= s on the same socket.", "Counter": "0,1,2,3", @@ -696,6 +909,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC0477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit in the L3 or were snooped from another core's caches on the sa= me socket.", "Counter": "0,1,2,3", @@ -736,6 +959,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, unless in Sub NUMA C= luster(SNC) Mode. In SNC Mode counts only those PMM accesses that are cont= rolled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.LOCAL_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100400477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, whether or not in Su= b NUMA Cluster(SNC) Mode. In SNC Mode counts PMM accesses that are control= led by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700C00477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F33000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by a cache on a remote socket where a snoop was sent= and data was returned (Modified or Not Modified).", "Counter": "0,1,2,3", @@ -766,6 +1019,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.REMOTE_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit a modified line in a distant L3 Cache or were snooped from a d= istant core's L1/L2 caches on this socket when the system is in SNC (sub-NU= MA cluster) mode.", "Counter": "0,1,2,3", @@ -786,6 +1049,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM on a distant memory controller of this socket= when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.SNC_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700800477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that hit in the L3 or= were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json b/too= ls/perf/pmu-events/arch/x86/icelakex/icx-metrics.json index 7bee03e532e4..a886a0cfee07 100644 --- a/tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json +++ b/tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json @@ -335,12 +335,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6) / (4 * tma_info_core_core_= clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -352,7 +352,7 @@ "MetricExpr": "34 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, @@ -375,12 +375,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -395,7 +395,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -403,7 +403,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store)) + tma_memory_bound * (tma_stor= e_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tm= a_store_bound)) * (tma_store_latency / (tma_store_latency + tma_false_shari= ng + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores)) + tma_memory_bound * (tma_stor= e_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tm= a_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_false_sharing = + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -411,22 +411,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_br= anches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma= _ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms /= (tma_mite + tma_dsb + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_br= anches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tm= a_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms /= (tma_dsb + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_mispredicts_resteers += tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_it= lb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ms)) + 10 *= tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts *= tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_= nukes + tma_core_bound * (tma_serializing_operation + tma_core_bound * RS_E= VENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0) / (tma_di= vider + tma_serializing_operation + tma_ports_utilization) + tma_microcode_= sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * (tma_as= sists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_clears_resteers + tma_= mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_= dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_mite + tma_ms)) + 10 *= tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mispredicts *= tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_= nukes + tma_core_bound * (tma_serializing_operation + tma_core_bound * RS_E= VENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0) / (tma_di= vider + tma_ports_utilization + tma_serializing_operation) + tma_microcode_= sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * (tma_as= sists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -434,7 +434,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -442,7 +442,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency + tma_streaming_stores - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -450,7 +450,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -462,17 +462,17 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions.", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES= / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_branch_instructions", @@ -494,8 +494,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -503,8 +503,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -512,24 +512,24 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -537,7 +537,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -545,33 +545,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((48 * tma_info_system_core_frequency - 4 * tma_inf= o_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (OCR.DEMAND= _DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DE= MAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) + (47.5 * tma_info_system_core_fr= equency - 4 * tma_info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSN= P_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tm= a_info_thread_clks", + "MetricExpr": "(44 * tma_info_system_core_frequency * (MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM * (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAN= D_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD)= )) + 43.5 * tma_info_system_core_frequency * MEM_LOAD_L3_HIT_RETIRED.XSNP_M= ISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_i= nfo_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -581,25 +581,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(47.5 * tma_info_system_core_frequency - 4 * tma_in= fo_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM * (1 - OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_= FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma= _info_thread_clks", + "MetricExpr": "43.5 * tma_info_system_core_frequency * (MEM_LOAD_L= 3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (1 - OCR.DEMAN= D_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT /= MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma= _remote_cache", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -608,7 +608,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -618,7 +618,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -628,7 +628,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -636,34 +636,34 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", - "MetricExpr": "(120 * tma_info_system_core_frequency * cpu@OCR.DEM= AND_RFO.L3_MISS\\,offcore_rsp\\=3D0x103b800002@ + 48 * tma_info_system_core= _frequency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", + "MetricExpr": "(120 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_MISS@offcore_rsp\\=3D0x103b800002@ + 48 * tma_info_system_core_frequ= ency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, @@ -683,7 +683,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -693,7 +693,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -711,7 +711,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -720,15 +720,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -736,7 +736,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -745,7 +745,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -754,7 +754,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -763,7 +763,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -772,7 +772,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -784,17 +784,17 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D0x1@) / IDQ.MITE_UOPS", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D1@) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -802,8 +802,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -815,28 +815,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -865,7 +865,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -874,7 +874,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -883,10 +883,11 @@ { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -947,11 +948,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -964,20 +965,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -1013,7 +1014,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -1031,7 +1032,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -1039,7 +1040,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -1047,7 +1048,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -1055,7 +1056,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -1063,7 +1064,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -1071,7 +1072,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1126,7 +1127,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 11", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1257,7 +1258,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1319,8 +1320,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1341,12 +1342,12 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1401,14 +1402,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1429,11 +1429,11 @@ "MetricExpr": "UNC_CHA_RxC_IRQ1_REJECT.PA_MATCH / UNC_CHA_CLOCKTIC= KS", "MetricGroup": "LockCont;MemOffcore;Server;SoC", "MetricName": "tma_info_system_mem_irq_duplicate_address", - "MetricThreshold": "(tma_info_system_mem_irq_duplicate_address > 0= .1)" + "MetricThreshold": "tma_info_system_mem_irq_duplicate_address > 0.= 1" }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1463,7 +1463,7 @@ "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_= clks", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1471,7 +1471,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1479,7 +1479,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1513,7 +1513,7 @@ "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1522,15 +1522,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1540,13 +1539,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1562,14 +1561,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 7.5" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1577,8 +1576,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1586,7 +1585,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1595,7 +1594,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1605,7 +1604,7 @@ "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + L1D_PEND_MISS.FB_FULL_PERIODS)= * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1614,7 +1613,7 @@ "MetricExpr": "4 * tma_info_system_core_frequency * MEM_LOAD_RETIR= ED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / = tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1624,17 +1623,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(23 * tma_info_system_core_frequency - 4 * tma_info= _system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETIRED.= FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "19 * tma_info_system_core_frequency * (MEM_LOAD_RET= IRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2))= / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1642,18 +1641,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1670,7 +1669,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1678,39 +1677,39 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from local memory", - "MetricExpr": "(66.5 * tma_info_system_core_frequency - 23 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "43.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _MISS_RETIRED.LOCAL_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1720,7 +1719,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1736,10 +1735,10 @@ }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1748,7 +1747,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1759,11 +1758,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", @@ -1785,7 +1784,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1800,24 +1799,24 @@ }, { "BriefDescription": "This metric represents fraction of cycles whe= re (only) 4 uops were delivered by the MITE pipeline", - "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D0x4@ - cpu@IDQ.MITE_= UOPS\\,cmask\\=3D0x5@) / tma_info_thread_clks", + "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D4@ - cpu@IDQ.MITE_UO= PS\\,cmask\\=3D5@) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_gr= oup", "MetricName": "tma_mite_4wide", - "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_f= etch_bandwidth > 0.2", + "MetricThreshold": "tma_mite_4wide > 0.05 & (tma_mite > 0.1 & tma_= fetch_bandwidth > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D0x1@ / tma_info_core_co= re_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D1@ / tma_info_core_core= _clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1828,7 +1827,7 @@ "MetricExpr": "3 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1837,7 +1836,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -1852,19 +1851,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1908,8 +1907,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1917,8 +1916,8 @@ "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=3D0x80@ / t= ma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1926,7 +1925,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1935,7 +1934,7 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= port_0, tma_port_1, tma_port_5, tma_port_6", "ScaleUnit": "100%" }, @@ -1944,32 +1943,32 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote cache in other socket= s including synchronizations issues", - "MetricExpr": "((120 * tma_info_system_core_frequency - 23 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + (120 * = tma_info_system_core_frequency - 23 * tma_info_system_core_frequency) * MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD= _RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(97 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_HITM + 97 * tma_info_system_core_frequency * MEM_LOAD_L= 3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRE= D.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote memory", - "MetricExpr": "(131 * tma_info_system_core_frequency - 23 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "108 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -1982,7 +1981,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1991,7 +1990,7 @@ "MetricExpr": "37 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUS= E_INST", "ScaleUnit": "100%" }, @@ -2001,7 +2000,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -2010,8 +2009,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -2019,7 +2018,7 @@ "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -2028,8 +2027,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -2038,8 +2037,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -2047,8 +2046,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 10 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -2065,7 +2064,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -2073,31 +2072,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2105,7 +2104,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -2114,7 +2113,7 @@ "MetricExpr": "10 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -2123,8 +2122,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/icelakex/memory.json b/tools/pe= rf/pmu-events/arch/x86/icelakex/memory.json index ec9577cce3ac..ca7f68f67463 100644 --- a/tools/perf/pmu-events/arch/x86/icelakex/memory.json +++ b/tools/perf/pmu-events/arch/x86/icelakex/memory.json @@ -113,6 +113,16 @@ "SampleAfterValue": "50021", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the local socket's L1, L= 2, or L3 caches.", "Counter": "0,1,2,3", @@ -133,6 +143,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -153,6 +193,46 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were n= ot supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -173,6 +253,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that were not supplied by the local so= cket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -193,6 +303,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that were supplied by DRAM attached to= this socket, unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts onl= y those DRAM accesses that are controlled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t missed the local socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -253,6 +373,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -283,6 +413,56 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x70C000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x731800477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.READS_TO_CORE.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that missed the local= socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -303,6 +483,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.WRITE_ESTIMATE.MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0xFBFF80822", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/icelakex/other.json b/tools/per= f/pmu-events/arch/x86/icelakex/other.json index 05b348d9c838..141cd30a30af 100644 --- a/tools/perf/pmu-events/arch/x86/icelakex/other.json +++ b/tools/perf/pmu-events/arch/x86/icelakex/other.json @@ -26,339 +26,6 @@ "SampleAfterValue": "200003", "UMask": "0x20" }, - { - "BriefDescription": "Hit snoop reply with data, line invalidated.", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.I_FWD_FE", - "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated: removed from this core's cache, after the dat= a is forwarded back to the requestor and indicating the data was found unmo= dified in the (FE) Forward or Exclusive State in this cores caches cache. = A single snoop response from the core counts on all hyperthreads of the cor= e.", - "SampleAfterValue": "1000003", - "UMask": "0x20" - }, - { - "BriefDescription": "HitM snoop reply with data, line invalidated.= ", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.I_FWD_M", - "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated: removed from this core's caches, after the da= ta is forwarded back to the requestor, and indicating the data was found mo= dified(M) in this cores caches cache (aka HitM response). A single snoop r= esponse from the core counts on all hyperthreads of the core.", - "SampleAfterValue": "1000003", - "UMask": "0x10" - }, - { - "BriefDescription": "Hit snoop reply without sending the data, lin= e invalidated.", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.I_HIT_FSE", - "PublicDescription": "Counts responses to snoops indicating the li= ne will now be (I)nvalidated in this core's caches without being forwarded = back to the requestor. The line was in Forward, Shared or Exclusive (FSE) s= tate in this cores caches. A single snoop response from the core counts on= all hyperthreads of the core.", - "SampleAfterValue": "1000003", - "UMask": "0x2" - }, - { - "BriefDescription": "Line not found snoop reply", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.MISS", - "PublicDescription": "Counts responses to snoops indicating that t= he data was not found (IHitI) in this core's caches. A single snoop respons= e from the core counts on all hyperthreads of the Core.", - "SampleAfterValue": "1000003", - "UMask": "0x1" - }, - { - "BriefDescription": "Hit snoop reply with data, line kept in Share= d state.", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.S_FWD_FE", - "PublicDescription": "Counts responses to snoops indicating the li= ne may be kept on this core in the (S)hared state, after the data is forwar= ded back to the requestor, initially the data was found in the cache in the= (FS) Forward or Shared state. A single snoop response from the core count= s on all hyperthreads of the core.", - "SampleAfterValue": "1000003", - "UMask": "0x40" - }, - { - "BriefDescription": "HitM snoop reply with data, line kept in Shar= ed state", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.S_FWD_M", - "PublicDescription": "Counts responses to snoops indicating the li= ne may be kept on this core in the (S)hared state, after the data is forwar= ded back to the requestor, initially the data was found in the cache in the= (M)odified state. A single snoop response from the core counts on all hyp= erthreads of the core.", - "SampleAfterValue": "1000003", - "UMask": "0x8" - }, - { - "BriefDescription": "Hit snoop reply without sending the data, lin= e kept in Shared state.", - "Counter": "0,1,2,3", - "EventCode": "0xef", - "EventName": "CORE_SNOOP_RESPONSE.S_HIT_FSE", - "PublicDescription": "Counts responses to snoops indicating the li= ne was kept on this core in the (S)hared state, and that the data was found= unmodified but not forwarded back to the requestor, initially the data was= found in the cache in the (FSE) Forward, Shared state or Exclusive state. = A single snoop response from the core counts on all hyperthreads of the co= re.", - "SampleAfterValue": "1000003", - "UMask": "0x4" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In SN= C Mode counts only those PMM accesses that are controlled by the close SNC = Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703C00001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM on a distant memory controller of this socket when the system is in S= NC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.SNC_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700800001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC0002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM attached to this socket, unless in Sub NUMA Cluster(SNC) Mod= e. In SNC Mode counts only those PMM accesses that are controlled by the c= lose SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.LOCAL_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703C00002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.REMOTE_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by PMM on a distant memory controller of this socket when the syste= m is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.SNC_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700800002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that were supplied by DRAM attached to= this socket, unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts onl= y those DRAM accesses that are controlled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch (which bring data to= L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10070", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L3.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x12380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L3.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x90002380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts full cacheline writes (ItoM) that were= not supplied by the local socket's L1, L2, or L3 caches and the cacheline = was homed in a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.ITOM.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x90000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that have any type of response.", "Counter": "0,1,2,3", @@ -369,126 +36,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC0477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, unless in Sub NUMA C= luster(SNC) Mode. In SNC Mode counts only those PMM accesses that are cont= rolled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.LOCAL_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100400477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x70C000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, whether or not in Su= b NUMA Cluster(SNC) Mode. In SNC Mode counts PMM accesses that are control= led by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700C00477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F33000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x731800477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.REMOTE_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM on a distant memory controller of this socket= when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.READS_TO_CORE.SNC_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700800477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -498,15 +45,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.WRITE_ESTIMATE.MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0xFBFF80822", - "SampleAfterValue": "100003", - "UMask": "0x1" } ] --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 14018376 for ; Wed, 26 Mar 2025 00:07:45 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947668; cv=none; b=OWu+/98tlNoguusam5McixPjuSn9kBTOWDSgc1N6TaztVk6HJ63KAaAq4t/vbb2IoCBDzE1ryrBUNxLh/kdroRKcCeL6pBzOlNnYq48BAMoaVbaf9KD4v9X05ERbPvkegdNy96KSxGkvYVJsYeAEPWj3npTUDNXbPJFA4LKM0s4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947668; c=relaxed/simple; bh=VCgSUoDyn8HKLqzmgh7KL88mTyxyID3ZHa2mEWl9iLg=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=PZNEoncN/nJYaAo65gALPUrfVnuUbR+XIsc8KQuB+kdRzlDy4cX6P5BTQlOCzTn+wbp+QGNvEoVogSeFpuJYNqbwajq/0FwIbf4A5Z5P5fKmK7Gq9rl3/z+aGR8ebshIQo5be5JI6Ucmp9v5F40HLQiUJKsP718Kcyt1Uhywuq0= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=EvrG82UN; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="EvrG82UN" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2241e7e3addso91365185ad.1 for ; Tue, 25 Mar 2025 17:07:45 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947664; x=1743552464; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=ni4SJHLrRP3Qlhbq7D+TPU+znQiai8CoM4wEG4fVdKI=; b=EvrG82UNObNX22YiHtVu4QLuOcplnSGOmeFEHhSgVv9UCCGBusj/Wzv0kYr5tw/ubO Xdav7zXrnVi9V8l0Epueh31WKap5UYUtKHFh06LbkrpH55XW+oD7Iw6oWNexNCWWPdVn fdIDOcubVVclu5WSxAXu4+29jRj3xPY+DGJNhmuzXh8D5N5SQdIM4+NPa2n3jE2hQVTm SACe+udcF2WaOhPvML74otOxnyc57VJMg1N/bElhjZJBLSbO9wK0ZrK4MdJ/MRzsnmEJ YbuvcpA+defj/3pJIsw0zqkS1gPz2bIMRdSBwyZDQqgPCqz0sWqPM41LQDMXMet2H5Z3 EC6w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947664; x=1743552464; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=ni4SJHLrRP3Qlhbq7D+TPU+znQiai8CoM4wEG4fVdKI=; b=ZN+nWq0slhSRTknrncs1sUSbUvouYe85LNsQIkkAII9GyJjid7q/8Aes7CWckiU7Vf PV+xDy9D6kstgQ/RLAAZ9ir52Kg/9Oz4DJZInudy8fsB/KckCzqeWASjVK+qxxRKgAQC Q/qv/ORCyKeY9EPtaSivY5qY0DnBraY46+tWNgHZ5M45kMZY3ENwh31QbqWfvZfxLFvN oJCWFCuuRCcuIrjgkdm2Eaa5+x8l3gQjeLRzeRK2OxtN6UFw8kn47XxvdRBNm2kL5ZKc N0iKD8OdQwKAFprZoLueOnXGWbW4qMMs6oxGTXYiRHCuKqwO9oYdtM5KFwA+R8Ss6bF6 3saw== X-Forwarded-Encrypted: i=1; AJvYcCUtQEKWx7y1nT94BFzA4Uc2/ZoOR19GUR26nfRoiba7SqBMsA4+TpvBTiNUERzeEmn6URTXoZvN6o9JPys=@vger.kernel.org X-Gm-Message-State: AOJu0YxXiy8LmWvG+SrPyWV8JqmRf4sW5X2Ot2td33IRJZkkQ+9g+E87 aJa0cMiwB1qhj1hJaWBziGhiYXIVc/3RXKnKfiSMliiDLMsjDLUpxoibn1RRHeUPFqyvhMcDyND PZvD/+Q== X-Google-Smtp-Source: AGHT+IE0yvE5R8zS++QQwXmgFe9oEWEr63eqK9HNjwtMO4jIRQ9T3a/sgvCXuGeOd9MwabBUfpz9fRzAV7Bp X-Received: from pffa21.prod.google.com ([2002:a62:bd15:0:b0:736:6224:4b5b]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:80e:b0:730:79bf:c893 with SMTP id d2e1a72fcca58-7390593d618mr25933695b3a.4.1742947664522; Tue, 25 Mar 2025 17:07:44 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:16 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-19-irogers@google.com> Subject: [PATCH v3 18/35] perf vendor events: Update ivybridge metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update TMA metrics from 4.8 to 5.02. Signed-off-by: Ian Rogers --- .../arch/x86/ivybridge/ivb-metrics.json | 76 +++++++++++++------ .../arch/x86/ivybridge/metricgroups.json | 5 ++ 2 files changed, 56 insertions(+), 25 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/ivybridge/ivb-metrics.json b/to= ols/perf/pmu-events/arch/x86/ivybridge/ivb-metrics.json index 77d37db98b70..de651ff9f846 100644 --- a/tools/perf/pmu-events/arch/x86/ivybridge/ivb-metrics.json +++ b/tools/perf/pmu-events/arch/x86/ivybridge/ivb-metrics.json @@ -151,7 +151,7 @@ "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "(60 * (MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM * (1= + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD= _UOPS_RETIRED.LLC_HIT + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_LLC_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MISS + M= EM_LOAD_UOPS_RETIRED.LLC_MISS))) + 43 * (MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP= _MISS * (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT = + MEM_LOAD_UOPS_RETIRED.LLC_HIT + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT + = MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSN= P_MISS + MEM_LOAD_UOPS_RETIRED.LLC_MISS)))) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop;TopdownL4;tma_L4_gr= oup;tma_issueSyncxn;tma_l3_bound_group", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", @@ -184,7 +184,7 @@ "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -236,7 +236,7 @@ { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", "MetricExpr": "60 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER= _CORE / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop;TopdownL4;tma_L4_gr= oup;tma_issueSyncxn;tma_store_bound_group", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", @@ -246,7 +246,7 @@ "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", @@ -305,7 +305,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -314,7 +314,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -334,7 +334,7 @@ "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences. ([ICL+] Note this may overcou= nt due to approximation using indirect events; [ADL+] .)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -346,7 +346,7 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", @@ -397,6 +397,12 @@ "MetricGroup": "Fed", "MetricName": "tma_info_frontend_ipunknown_branch" }, + { + "BriefDescription": "Taken Branches retired Per Cycle", + "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", + "MetricGroup": "Branches;FetchBW", + "MetricName": "tma_info_frontend_tbpc" + }, { "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", @@ -474,7 +480,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L1 data cache [GB / sec]", - "MetricExpr": "64 * L1D.REPLACEMENT / 1e9 / duration_time", + "MetricExpr": "64 * L1D.REPLACEMENT / 1e9 / tma_info_system_time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, @@ -486,7 +492,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L2 cache [GB / sec]", - "MetricExpr": "64 * L2_LINES_IN.ALL / 1e9 / duration_time", + "MetricExpr": "64 * L2_LINES_IN.ALL / 1e9 / tma_info_system_time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, @@ -504,7 +510,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L3 cache [GB / sec]", - "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1e9 / duration_time", + "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1e9 / tma_info_system= _time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, @@ -523,7 +529,7 @@ { "BriefDescription": "Average Latency for L2 cache miss demand Load= s", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCO= RE_REQUESTS.DEMAND_DATA_RD", - "MetricGroup": "Memory_Lat;Offcore", + "MetricGroup": "LockCont;Memory_Lat;Offcore", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { @@ -555,7 +561,7 @@ "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", + "BriefDescription": "", "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" @@ -568,7 +574,7 @@ }, { "BriefDescription": "Measured Average Core Frequency for unhalted = processors [GHz]", - "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / dur= ation_time", + "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / tma= _info_system_time", "MetricGroup": "Power;Summary", "MetricName": "tma_info_system_core_frequency" }, @@ -586,14 +592,14 @@ }, { "BriefDescription": "Average external Memory Bandwidth Use for rea= ds and writes [GB / sec]", - "MetricExpr": "64 * (UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_RE= QUESTS.ALL) / 1e6 / duration_time / 1e3", + "MetricExpr": "64 * (UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_RE= QUESTS.ALL) / 1e6 / tma_info_system_time / 1e3", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for re= ads and writes [GB / sec]. Related metrics: tma_fb_full, tma_mem_bandwidth,= tma_sq_full" }, { "BriefDescription": "Giga Floating Point Operations Per Second", - "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / duration_time", + "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. A= ggregate across all supported options of: FP precisions, scalar and vector = instructions, vector-width" @@ -618,6 +624,19 @@ "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, + { + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD= ", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" + }, + { + "BriefDescription": "Total package Power in Watts", + "MetricExpr": "power@energy\\-pkg@ * 15.6 / (tma_info_system_time = * 1e6)", + "MetricGroup": "Power;SoC", + "MetricName": "tma_info_system_power" + }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", "MetricExpr": "(1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / (CPU_CLK_= UNHALTED.REF_XCLK_ANY / 2) if #SMT_on else 0)", @@ -630,6 +649,13 @@ "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, + { + "BriefDescription": "Run duration time in seconds", + "MetricExpr": "duration_time", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" + }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", @@ -691,12 +717,12 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates how often the CPU was s= talled without loads missing the L1 data cache", + "BriefDescription": "This metric estimates how often the CPU was s= talled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.ST= ALLS_LDM_PENDING) - CYCLE_ACTIVITY.STALLS_L1D_PENDING) / tma_info_thread_cl= ks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 data cache. The L1 data cache typical= ly has the shortest latency. However; in certain cases like loads blocked = on older stores; a load might suffer due to high latency even though it is = being satisfied by the L1. Another example is loads who miss in the TLB. Th= ese cases are characterized by execution unit stalls; while some non-comple= ted demand load lives in the machine without having that demand load missin= g the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS;MEM_LOAD_UOPS_= RETIRED.HIT_LFB_PS. Related metrics: tma_clears_resteers, tma_machine_clear= s, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -761,7 +787,7 @@ "BriefDescription": "This metric represents fraction of cycles the= CPU spent handling cache misses due to lock operations", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", - "MetricGroup": "Offcore;TopdownL4;tma_L4_group;tma_issueRFO;tma_l1= _bound_group", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", @@ -781,7 +807,7 @@ { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", @@ -840,7 +866,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vecto= r_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -849,7 +875,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tm= a_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -949,7 +975,7 @@ "MetricExpr": "13 * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", - "MetricThreshold": "tma_split_loads > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, @@ -965,7 +991,7 @@ { "BriefDescription": "This metric measures fraction of cycles where= the Super Queue (SQ) was full taking into account all request-types and bo= th hardware SMT threads (Logical Processors)", "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", @@ -993,7 +1019,7 @@ "BriefDescription": "This metric estimates fraction of cycles the = CPU spent handling L1D store misses", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= issueRFO;tma_issueSL;tma_store_bound_group", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", diff --git a/tools/perf/pmu-events/arch/x86/ivybridge/metricgroups.json b/t= ools/perf/pmu-events/arch/x86/ivybridge/metricgroups.json index 4193c90c3459..0863375bdead 100644 --- a/tools/perf/pmu-events/arch/x86/ivybridge/metricgroups.json +++ b/tools/perf/pmu-events/arch/x86/ivybridge/metricgroups.json @@ -9,6 +9,7 @@ "BvCB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvFB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvIO": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", + "BvMB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvML": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMP": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMS": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", @@ -34,6 +35,7 @@ "InsType": "Grouping from Top-down Microarchitecture Analysis Metrics = spreadsheet", "L2Evicts": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "LSD": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", + "LockCont": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "MachineClears": "Grouping from Top-down Microarchitecture Analysis Me= trics spreadsheet", "Machine_Clears": "Grouping from Top-down Microarchitecture Analysis M= etrics spreadsheet", "Mem": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -51,6 +53,7 @@ "Pipeline": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "PortsUtil": "Grouping from Top-down Microarchitecture Analysis Metric= s spreadsheet", "Power": "Grouping from Top-down Microarchitecture Analysis Metrics sp= readsheet", + "Prefetches": "Grouping from Top-down Microarchitecture Analysis Metri= cs spreadsheet", "Ret": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", "Retire": "Grouping from Top-down Microarchitecture Analysis Metrics s= preadsheet", "SMT": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -78,6 +81,7 @@ "tma_bad_speculation_group": "Metrics contributing to tma_bad_speculat= ion category", "tma_branch_resteers_group": "Metrics contributing to tma_branch_reste= ers category", "tma_core_bound_group": "Metrics contributing to tma_core_bound catego= ry", + "tma_divider_group": "Metrics contributing to tma_divider category", "tma_dram_bound_group": "Metrics contributing to tma_dram_bound catego= ry", "tma_dtlb_load_group": "Metrics contributing to tma_dtlb_load category= ", "tma_dtlb_store_group": "Metrics contributing to tma_dtlb_store catego= ry", @@ -103,6 +107,7 @@ "tma_issueSpSt": "Metrics related by the issue $issueSpSt", "tma_issueSyncxn": "Metrics related by the issue $issueSyncxn", "tma_issueTLB": "Metrics related by the issue $issueTLB", + "tma_itlb_misses_group": "Metrics contributing to tma_itlb_misses cate= gory", "tma_l1_bound_group": "Metrics contributing to tma_l1_bound category", "tma_l3_bound_group": "Metrics contributing to tma_l3_bound category", "tma_light_operations_group": "Metrics contributing to tma_light_opera= tions category", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oa1-f73.google.com (mail-oa1-f73.google.com [209.85.160.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 233CB13D539 for ; Wed, 26 Mar 2025 00:07:53 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.160.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947677; cv=none; b=HAvM7wfbmm8BQXMbUWDteM/tixjutACy8O95NKMwjezkJhfVdiV8y98qfyW6800+FBCJUBWwKtB0ZQPHe2DGtHx7rF5Js2a90TLGy93AOXJodxQ8D4w1Qcl1rclm0gvIelMAHn2ZdmKXG4WNhCZbsJrdDV7NdYNohKBXmbabYaQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947677; c=relaxed/simple; bh=CnuQSxx5Yu1wyHO5RngC24AJ1X96YzXrl2hjW0dNQbI=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=BF9QGpT0JVKpAA73NUvoAPixbyKwuOG/5Y1JXg0KH7ZBoQ8ejNRaTV7R3o4Icr1uFNaGnrVR71eafXZHCrIUHtElFegKMVBewpuPrD3+EDwoGza4mj+t/YNrexYVfiUZOV3hF1543iX91AZhBU0JCbaEwPgMs9uGOVsnch2hqE0= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=HWf01fbB; arc=none smtp.client-ip=209.85.160.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="HWf01fbB" Received: by mail-oa1-f73.google.com with SMTP id 586e51a60fabf-2c2fc98f199so5069988fac.1 for ; Tue, 25 Mar 2025 17:07:53 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947673; x=1743552473; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=s+cdTdHDpZF3mMROwANPX8DYhSkxmnBHZWoA/0TRdpI=; b=HWf01fbBreyKqPttDecSnS3Ot0N+lGVhPHgyZmyuPOgezYPtQPvYrE4YELMLnV6scl ipVYv34uZCgK76I35uAmUxogUd2MJJRjtTjoSFddqF9jsH8BGKH2yials+HbDff5k/Z5 y6g5UW/V3JBemGY9VVeh9zzsiLBsAyGR+9blh+5XfjSMK1kvs23yXpwvF/MMkwBR/+LP yI1AqR4jd+0n1FZf0MHH6LzBUAeo6NUFJEsL2pfBsjAApKOuDHCNySKloRWPVu8vCw2v tXX/jH1IM6DdQfd2R4heww8zRWTMl2V1fxGa6oqT3dW+I8W6AkEm3RJQo7He5egiU4CB HluQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947673; x=1743552473; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=s+cdTdHDpZF3mMROwANPX8DYhSkxmnBHZWoA/0TRdpI=; b=Ypx4qrNY5xMBXVmv/PswzvkHBibLFUrFblgD82n89k8n9hUEpVwMUYigQ7KJMfo0xf vlZIqiFJhvXqz/MLQjReqxf0ydELUI1culXDvyUzGzJzFrliCLuINgEGxitBk5L0oAgo hjw0ZrDWNyAwEUb4ymci70T0XRxqwQsDe7GzaTyimOwZUKNyDwS2Unl50GItoIWTMTZs x9Gk1n6Suc6YzstHapoC+e8xl2dl3r8ryEOW3hte0Nk0Pb1hda6aWBAEQs4CCtM2fDsJ NwuoazUomD8aFijBFv/yAszEZ5cTgi/2T5XD2zOZmx6Imuj5hpK9bi9IIGokaG8+pUCR zDTA== X-Forwarded-Encrypted: i=1; AJvYcCWth1KTjiqGeXLbWWDa46dmrye7eh6uk5xqJrUsITyOAREtGgtbLmDRe69odF8DFzqVXPoSsbiKdrOMnHk=@vger.kernel.org X-Gm-Message-State: AOJu0YwdbAs4rHKbKtMvihRa3/gBhl9VIzwC7EHN5O7+0T08uYrkP5ZJ KT/jacBGdE2x7SH3Kt4RlcVYFrqJHBBSfn2LhRXuy3pKtmhKUMXe7DaYcNu83keY3MeIGlg34ud puy/iwA== X-Google-Smtp-Source: AGHT+IGlpb+dFAh2ypRGPFaDSawGEO3G27OdyvftMmycas0mMZis5PjP5cE7Dqtz7mS85jlbPE6AFS8BOcIL X-Received: from oacpi3.prod.google.com ([2002:a05:6871:d103:b0:2b8:443e:cfbd]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6870:40ca:b0:2c1:c10e:3230 with SMTP id 586e51a60fabf-2c780351832mr10080063fac.18.1742947673244; Tue, 25 Mar 2025 17:07:53 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:17 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-20-irogers@google.com> Subject: [PATCH v3 19/35] perf vendor events: Update ivytown metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update TMA metrics from 4.8 to 5.02. Signed-off-by: Ian Rogers --- .../arch/x86/ivytown/ivt-metrics.json | 80 ++++++++++++------- .../arch/x86/ivytown/metricgroups.json | 5 ++ 2 files changed, 58 insertions(+), 27 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/ivytown/ivt-metrics.json b/tool= s/perf/pmu-events/arch/x86/ivytown/ivt-metrics.json index 8fe0512c938f..714d5e6d21e7 100644 --- a/tools/perf/pmu-events/arch/x86/ivytown/ivt-metrics.json +++ b/tools/perf/pmu-events/arch/x86/ivytown/ivt-metrics.json @@ -151,7 +151,7 @@ "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "(60 * (MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM * (1= + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM_LOAD= _UOPS_RETIRED.LLC_HIT + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT + MEM_LOAD_U= OPS_LLC_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MISS + M= EM_LOAD_UOPS_LLC_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_LLC_MISS_RETIRED.R= EMOTE_DRAM + MEM_LOAD_UOPS_LLC_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOPS_LLC= _MISS_RETIRED.REMOTE_FWD))) + 43 * (MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MISS= * (1 + MEM_LOAD_UOPS_RETIRED.HIT_LFB / (MEM_LOAD_UOPS_RETIRED.L2_HIT + MEM= _LOAD_UOPS_RETIRED.LLC_HIT + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT + MEM_L= OAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM + MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MIS= S + MEM_LOAD_UOPS_LLC_MISS_RETIRED.LOCAL_DRAM + MEM_LOAD_UOPS_LLC_MISS_RETI= RED.REMOTE_DRAM + MEM_LOAD_UOPS_LLC_MISS_RETIRED.REMOTE_HITM + MEM_LOAD_UOP= S_LLC_MISS_RETIRED.REMOTE_FWD)))) / tma_info_thread_clks", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop;TopdownL4;tma_L4_gr= oup;tma_issueSyncxn;tma_l3_bound_group", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_data_sharing, tma_false_sharing, tma_machine_clears, tma= _remote_cache", @@ -184,7 +184,7 @@ "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -236,7 +236,7 @@ { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", "MetricExpr": "(200 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_MISS.REMOTE_= HITM + 60 * OFFCORE_RESPONSE.DEMAND_RFO.LLC_HIT.HITM_OTHER_CORE) / tma_info= _thread_clks", - "MetricGroup": "BvMS;DataSharing;Offcore;Snoop;TopdownL4;tma_L4_gr= oup;tma_issueSyncxn;tma_store_bound_group", + "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_contested_accesses, tma_data_sharing, tma_machine_clears, tma= _remote_cache", @@ -246,7 +246,7 @@ "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", + "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_info_system_dram_bw_use, tma_mem_ba= ndwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", @@ -305,7 +305,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -314,7 +314,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -334,7 +334,7 @@ "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences. ([ICL+] Note this may overcou= nt due to approximation using indirect events; [ADL+] .)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -346,7 +346,7 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "Instructions per retired mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", @@ -397,6 +397,12 @@ "MetricGroup": "Fed", "MetricName": "tma_info_frontend_ipunknown_branch" }, + { + "BriefDescription": "Taken Branches retired Per Cycle", + "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", + "MetricGroup": "Branches;FetchBW", + "MetricName": "tma_info_frontend_tbpc" + }, { "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", @@ -474,7 +480,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L1 data cache [GB / sec]", - "MetricExpr": "64 * L1D.REPLACEMENT / 1e9 / duration_time", + "MetricExpr": "64 * L1D.REPLACEMENT / 1e9 / tma_info_system_time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l1d_cache_fill_bw" }, @@ -486,7 +492,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L2 cache [GB / sec]", - "MetricExpr": "64 * L2_LINES_IN.ALL / 1e9 / duration_time", + "MetricExpr": "64 * L2_LINES_IN.ALL / 1e9 / tma_info_system_time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l2_cache_fill_bw" }, @@ -504,7 +510,7 @@ }, { "BriefDescription": "Average per-thread data fill bandwidth to the= L3 cache [GB / sec]", - "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1e9 / duration_time", + "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1e9 / tma_info_system= _time", "MetricGroup": "Mem;MemoryBW", "MetricName": "tma_info_memory_l3_cache_fill_bw" }, @@ -523,7 +529,7 @@ { "BriefDescription": "Average Latency for L2 cache miss demand Load= s", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / OFFCO= RE_REQUESTS.DEMAND_DATA_RD", - "MetricGroup": "Memory_Lat;Offcore", + "MetricGroup": "LockCont;Memory_Lat;Offcore", "MetricName": "tma_info_memory_latency_load_l2_miss_latency" }, { @@ -555,7 +561,7 @@ "MetricThreshold": "tma_info_memory_tlb_page_walks_utilization > 0= .5" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", + "BriefDescription": "", "MetricExpr": "UOPS_EXECUTED.THREAD / (cpu@UOPS_EXECUTED.CORE\\,cm= ask\\=3D1@ / 2 if #SMT_on else UOPS_EXECUTED.CYCLES_GE_1_UOP_EXEC)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" @@ -568,7 +574,7 @@ }, { "BriefDescription": "Measured Average Core Frequency for unhalted = processors [GHz]", - "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / dur= ation_time", + "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / tma= _info_system_time", "MetricGroup": "Power;Summary", "MetricName": "tma_info_system_core_frequency" }, @@ -586,14 +592,14 @@ }, { "BriefDescription": "Average external Memory Bandwidth Use for rea= ds and writes [GB / sec]", - "MetricExpr": "64 * (UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) / 1e= 9 / duration_time", + "MetricExpr": "64 * (UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) / 1e= 9 / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for re= ads and writes [GB / sec]. Related metrics: tma_fb_full, tma_mem_bandwidth,= tma_sq_full" }, { "BriefDescription": "Giga Floating Point Operations Per Second", - "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / duration_time", + "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. A= ggregate across all supported options of: FP precisions, scalar and vector = instructions, vector-width" @@ -627,11 +633,24 @@ }, { "BriefDescription": "Average latency of data read request to exter= nal memory (in nanoseconds)", - "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / duration_time)", + "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / tma_info_system_time)", "MetricGroup": "Mem;MemoryLat;SoC", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetche= s. ([RKL+]memory-controller only)" }, + { + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD= ", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" + }, + { + "BriefDescription": "Total package Power in Watts", + "MetricExpr": "(power@energy\\-pkg@ + power@energy\\-ram@) * 15.6 = / (duration_time * 1e6)", + "MetricGroup": "Power;SoC", + "MetricName": "tma_info_system_power" + }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", "MetricExpr": "(1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / (CPU_CLK_= UNHALTED.REF_XCLK_ANY / 2) if #SMT_on else 0)", @@ -644,6 +663,13 @@ "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, + { + "BriefDescription": "Run duration time in seconds", + "MetricExpr": "duration_time", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" + }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", @@ -652,7 +678,7 @@ }, { "BriefDescription": "Measured Average Uncore Frequency for the SoC= [GHz]", - "MetricExpr": "tma_info_system_socket_clks / 1e9 / duration_time", + "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system= _time", "MetricGroup": "SoC", "MetricName": "tma_info_system_uncore_frequency" }, @@ -711,12 +737,12 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates how often the CPU was s= talled without loads missing the L1 data cache", + "BriefDescription": "This metric estimates how often the CPU was s= talled without loads missing the L1 Data (L1D) cache", "MetricExpr": "max((min(CPU_CLK_UNHALTED.THREAD, CYCLE_ACTIVITY.ST= ALLS_LDM_PENDING) - CYCLE_ACTIVITY.STALLS_L1D_PENDING) / tma_info_thread_cl= ks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", - "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 data cache. The L1 data cache typical= ly has the shortest latency. However; in certain cases like loads blocked = on older stores; a load might suffer due to high latency even though it is = being satisfied by the L1. Another example is loads who miss in the TLB. Th= ese cases are characterized by execution unit stalls; while some non-comple= ted demand load lives in the machine without having that demand load missin= g the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS;MEM_LOAD_UOPS_= RETIRED.HIT_LFB_PS. Related metrics: tma_clears_resteers, tma_machine_clear= s, tma_microcode_sequencer, tma_ms_switches, tma_ports_utilized_1", + "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS. Related me= trics: tma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tm= a_ms_switches, tma_ports_utilized_1", "ScaleUnit": "100%" }, { @@ -790,7 +816,7 @@ "BriefDescription": "This metric represents fraction of cycles the= CPU spent handling cache misses due to lock operations", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "MEM_UOPS_RETIRED.LOCK_LOADS / MEM_UOPS_RETIRED.ALL_= STORES * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTANDING.CYCLES_W= ITH_DEMAND_RFO) / tma_info_thread_clks", - "MetricGroup": "Offcore;TopdownL4;tma_L4_group;tma_issueRFO;tma_l1= _bound_group", + "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_UOPS_RETIRED.LOCK_LOA= DS_PS. Related metrics: tma_store_latency", @@ -810,7 +836,7 @@ { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_fb_full, tma_info_system_dram_bw_u= se, tma_sq_full", @@ -869,7 +895,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vecto= r_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -878,7 +904,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tm= a_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -997,7 +1023,7 @@ "MetricExpr": "13 * LD_BLOCKS.NO_SR / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", - "MetricThreshold": "tma_split_loads > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "MetricThreshold": "tma_split_loads > 0.3", "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_UOPS_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, @@ -1013,7 +1039,7 @@ { "BriefDescription": "This metric measures fraction of cycles where= the Super Queue (SQ) was full taking into account all request-types and bo= th hardware SMT threads (Logical Processors)", "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_fb_full= , tma_info_system_dram_bw_use, tma_mem_bandwidth", @@ -1041,7 +1067,7 @@ "BriefDescription": "This metric estimates fraction of cycles the = CPU spent handling L1D store misses", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_UOPS_RETIRED.LOCK_= LOADS / MEM_UOPS_RETIRED.ALL_STORES) + (1 - MEM_UOPS_RETIRED.LOCK_LOADS / M= EM_UOPS_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", - "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= issueRFO;tma_issueSL;tma_store_bound_group", + "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", diff --git a/tools/perf/pmu-events/arch/x86/ivytown/metricgroups.json b/too= ls/perf/pmu-events/arch/x86/ivytown/metricgroups.json index 4193c90c3459..0863375bdead 100644 --- a/tools/perf/pmu-events/arch/x86/ivytown/metricgroups.json +++ b/tools/perf/pmu-events/arch/x86/ivytown/metricgroups.json @@ -9,6 +9,7 @@ "BvCB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvFB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvIO": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", + "BvMB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvML": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMP": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMS": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", @@ -34,6 +35,7 @@ "InsType": "Grouping from Top-down Microarchitecture Analysis Metrics = spreadsheet", "L2Evicts": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "LSD": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", + "LockCont": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "MachineClears": "Grouping from Top-down Microarchitecture Analysis Me= trics spreadsheet", "Machine_Clears": "Grouping from Top-down Microarchitecture Analysis M= etrics spreadsheet", "Mem": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -51,6 +53,7 @@ "Pipeline": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "PortsUtil": "Grouping from Top-down Microarchitecture Analysis Metric= s spreadsheet", "Power": "Grouping from Top-down Microarchitecture Analysis Metrics sp= readsheet", + "Prefetches": "Grouping from Top-down Microarchitecture Analysis Metri= cs spreadsheet", "Ret": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", "Retire": "Grouping from Top-down Microarchitecture Analysis Metrics s= preadsheet", "SMT": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -78,6 +81,7 @@ "tma_bad_speculation_group": "Metrics contributing to tma_bad_speculat= ion category", "tma_branch_resteers_group": "Metrics contributing to tma_branch_reste= ers category", "tma_core_bound_group": "Metrics contributing to tma_core_bound catego= ry", + "tma_divider_group": "Metrics contributing to tma_divider category", "tma_dram_bound_group": "Metrics contributing to tma_dram_bound catego= ry", "tma_dtlb_load_group": "Metrics contributing to tma_dtlb_load category= ", "tma_dtlb_store_group": "Metrics contributing to tma_dtlb_store catego= ry", @@ -103,6 +107,7 @@ "tma_issueSpSt": "Metrics related by the issue $issueSpSt", "tma_issueSyncxn": "Metrics related by the issue $issueSyncxn", "tma_issueTLB": "Metrics related by the issue $issueTLB", + "tma_itlb_misses_group": "Metrics contributing to tma_itlb_misses cate= gory", "tma_l1_bound_group": "Metrics contributing to tma_l1_bound category", "tma_l3_bound_group": "Metrics contributing to tma_l3_bound category", "tma_light_operations_group": "Metrics contributing to tma_light_opera= tions category", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f201.google.com (mail-pl1-f201.google.com [209.85.214.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 4A1CD1494A8 for ; Wed, 26 Mar 2025 00:08:02 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.201 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947684; cv=none; b=PiScqvXJ5qC4xjMSiUXHrjVwkVs16qB9iOTOimWnccNFhIbbxoBGFtzpt+SxW4ypFMiPqAnnKqqOtLMpYC0c54o7x8jLGiBg0Eyoyaa6mTIyY0V2cnvKna2fvI3GASF+DwcaTH/avpeh90+T1c+bycoak0ZMhEFrLpYVjDg/V7Y= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947684; c=relaxed/simple; bh=bWiq+5MyDXTElYieLM6C7dh8a0sGhdsKcE3TX6IvQ64=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=HJDORs4EQOLnBRHs6/yglCqZFpqXe5Oea2Fs839dtuV0Uyz5qTp9IqNUOIOvYZdBaNig6l8e9kiFKwJvO7dP66OJGBLiCRQEDskVP/LVQ0ssikhWgCWqxdplef7afen2HQnJfTZfxbw5Ic4QcZkZZ3Cp5CZm3uv9sMiedRVcd80= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=wQBwA1yy; arc=none smtp.client-ip=209.85.214.201 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="wQBwA1yy" Received: by mail-pl1-f201.google.com with SMTP id d9443c01a7336-227ea16b03dso21105185ad.3 for ; Tue, 25 Mar 2025 17:08:02 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947682; x=1743552482; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=CC4DlL0ippkQfeYw3fEnFY+uPfIQ3TXIJOTrdzzBovo=; b=wQBwA1yyDmByFz65VT0Jh++HHitihRkuWYOCmK8m7BBaGwM/2wrSmwNnAsGrsLz0VR 1lZwJgqFXRySFgOvCoCJ4VhoTmELCpAto9HRKq8wZRc/b0Z3f/MQKKjqlSasoQexZ1PU z4/sHBBAk4asn75awSzQD1QNSl9VF1DXfa7lsbP8y2ziB5ZVCBKmzVobKbCAwj9XjFpQ r3jWAcUX2fngrc/NndflWu+v8rit4t2r2VwGyt8wuvrPT1iFQQpPAkCUscdSHuEKSuTg MuMqAwccMr0pLXz84VqCzijlQ5uIkLkte10oxb18Q6aVxuvycnTBOoIbWU7V7FhdZxpc dPrg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947682; x=1743552482; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=CC4DlL0ippkQfeYw3fEnFY+uPfIQ3TXIJOTrdzzBovo=; b=iqhLX7dY8I1cvOJnfttYou7XeuvAbtu3lx66Epaje6cOcZs1ELVidzBpPWiInNfHzx GxeQWMp2/Dnx2g8DOuSiBY858Rw1/IQprFikR38axdsnIpXiJGUvRkMTJQyHFeVcLMq5 h06nl5lhYOH+NMOKufZnLqs6bSeng5UZ/K6X1Md6XRiLpyUQltMsI5K1wnx49KgxHxlW Msxca/pn0okdhHRaO+B/UsmgXjVY9vHiZVDGR9K6Nh/K4T8GuPWU/msvx801JYrdsNbg bsJ1Ih/kVkHaafaoVlyB+hkfODf2piL7QvpUgrHAQvUUhDZB4CWVE8YgvExDdreC+Rpm 4Prg== X-Forwarded-Encrypted: i=1; AJvYcCXv3hEQSwJZh59Grixjcu7MJhdON8sVvfXA/AFbX1FE7CpMUTa1nezo/JQ2wvDmKyYCnzTL+L22dFDVT4w=@vger.kernel.org X-Gm-Message-State: AOJu0Yy1TP9GGthlxDdYt51+8E6RzdNBSmQrcryFidPeJvwODN1vZs5q MLXycPeTYrtGhTn7OhOuoKj6BlrHlfu1rvefPNu3hs/jbE9gHBW7RIJeK9sCNsmI7JX2bYxyq7v TU8vcfQ== X-Google-Smtp-Source: AGHT+IEqZec9MjahieD+WpQHzqgDKXtvSBbtGhEstSJJO3LzKYO5EJWIooXs5Y+dxXz322Iu5zE/nZTj2Wt1 X-Received: from pfbln22.prod.google.com ([2002:a05:6a00:3cd6:b0:737:79:e096]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:1409:b0:737:e73:f64b with SMTP id d2e1a72fcca58-7390593b7ffmr27333570b3a.1.1742947681731; Tue, 25 Mar 2025 17:08:01 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:18 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-21-irogers@google.com> Subject: [PATCH v3 20/35] perf vendor events: Update jaketown metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update TMA metrics from 4.8 to 5.02. Move INSTS_WRITTEN_TO_IQ.INSTS to the frontend topic. Signed-off-by: Ian Rogers --- .../arch/x86/jaketown/frontend.json | 8 ++++ .../arch/x86/jaketown/jkt-metrics.json | 40 ++++++++++++++----- .../arch/x86/jaketown/metricgroups.json | 5 +++ .../pmu-events/arch/x86/jaketown/other.json | 8 ---- 4 files changed, 43 insertions(+), 18 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/jaketown/frontend.json b/tools/= perf/pmu-events/arch/x86/jaketown/frontend.json index 3cb468da7011..97e7760aeb26 100644 --- a/tools/perf/pmu-events/arch/x86/jaketown/frontend.json +++ b/tools/perf/pmu-events/arch/x86/jaketown/frontend.json @@ -278,5 +278,13 @@ "EventName": "IDQ_UOPS_NOT_DELIVERED.CYCLES_LE_3_UOP_DELIV.CORE", "SampleAfterValue": "2000003", "UMask": "0x1" + }, + { + "BriefDescription": "Valid instructions written to IQ per cycle.", + "Counter": "0,1,2,3", + "EventCode": "0x17", + "EventName": "INSTS_WRITTEN_TO_IQ.INSTS", + "SampleAfterValue": "2000003", + "UMask": "0x1" } ] diff --git a/tools/perf/pmu-events/arch/x86/jaketown/jkt-metrics.json b/too= ls/perf/pmu-events/arch/x86/jaketown/jkt-metrics.json index f8c18741b360..6f636ea0f216 100644 --- a/tools/perf/pmu-events/arch/x86/jaketown/jkt-metrics.json +++ b/tools/perf/pmu-events/arch/x86/jaketown/jkt-metrics.json @@ -127,7 +127,7 @@ "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -211,7 +211,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_256b, tma_fp_vector_512b, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_6, tma_ports= _utilized_2", "ScaleUnit": "100%" }, { @@ -220,7 +220,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_512b, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_6, tma_ports= _utilized_2", "ScaleUnit": "100%" }, { @@ -240,7 +240,7 @@ "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences. ([ICL+] Note this may overcou= nt due to approximation using indirect events; [ADL+] .)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -275,6 +275,12 @@ "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & tma_inf= o_thread_ipc / 4 > 0.35", "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_lcp" }, + { + "BriefDescription": "Taken Branches retired Per Cycle", + "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", + "MetricGroup": "Branches;FetchBW", + "MetricName": "tma_info_frontend_tbpc" + }, { "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", @@ -290,7 +296,7 @@ }, { "BriefDescription": "Measured Average Core Frequency for unhalted = processors [GHz]", - "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / dur= ation_time", + "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / tma= _info_system_time", "MetricGroup": "Power;Summary", "MetricName": "tma_info_system_core_frequency" }, @@ -308,14 +314,14 @@ }, { "BriefDescription": "Average external Memory Bandwidth Use for rea= ds and writes [GB / sec]", - "MetricExpr": "64 * (UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) / 1e= 9 / duration_time", + "MetricExpr": "64 * (UNC_M_CAS_COUNT.RD + UNC_M_CAS_COUNT.WR) / 1e= 9 / tma_info_system_time", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for re= ads and writes [GB / sec]. Related metrics: tma_mem_bandwidth" }, { "BriefDescription": "Giga Floating Point Operations Per Second", - "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / duration_time", + "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. A= ggregate across all supported options of: FP precisions, scalar and vector = instructions, vector-width" @@ -349,11 +355,18 @@ }, { "BriefDescription": "Average latency of data read request to exter= nal memory (in nanoseconds)", - "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / duration_time)", + "MetricExpr": "1e9 * (UNC_C_TOR_OCCUPANCY.MISS_OPCODE@filter_opc\\= =3D0x182@ / UNC_C_TOR_INSERTS.MISS_OPCODE@filter_opc\\=3D0x182@) / (tma_inf= o_system_socket_clks / tma_info_system_time)", "MetricGroup": "Mem;MemoryLat;SoC", "MetricName": "tma_info_system_mem_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetche= s. ([RKL+]memory-controller only)" }, + { + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD= ", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" + }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", "MetricExpr": "(1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / (CPU_CLK_= UNHALTED.REF_XCLK_ANY / 2) if #SMT_on else 0)", @@ -366,6 +379,13 @@ "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, + { + "BriefDescription": "Run duration time in seconds", + "MetricExpr": "duration_time", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" + }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", @@ -374,7 +394,7 @@ }, { "BriefDescription": "Measured Average Uncore Frequency for the SoC= [GHz]", - "MetricExpr": "tma_info_system_socket_clks / 1e9 / duration_time", + "MetricExpr": "tma_info_system_socket_clks / 1e9 / tma_info_system= _time", "MetricGroup": "SoC", "MetricName": "tma_info_system_uncore_frequency" }, @@ -468,7 +488,7 @@ { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_info_system_dram_bw_use", diff --git a/tools/perf/pmu-events/arch/x86/jaketown/metricgroups.json b/to= ols/perf/pmu-events/arch/x86/jaketown/metricgroups.json index 7dc7eb0d3dd3..eb8fbd14138a 100644 --- a/tools/perf/pmu-events/arch/x86/jaketown/metricgroups.json +++ b/tools/perf/pmu-events/arch/x86/jaketown/metricgroups.json @@ -9,6 +9,7 @@ "BvCB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvFB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvIO": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", + "BvMB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvML": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMP": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMS": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", @@ -33,6 +34,7 @@ "InsType": "Grouping from Top-down Microarchitecture Analysis Metrics = spreadsheet", "L2Evicts": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "LSD": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", + "LockCont": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "MachineClears": "Grouping from Top-down Microarchitecture Analysis Me= trics spreadsheet", "Machine_Clears": "Grouping from Top-down Microarchitecture Analysis M= etrics spreadsheet", "Mem": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -48,6 +50,7 @@ "Pipeline": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "PortsUtil": "Grouping from Top-down Microarchitecture Analysis Metric= s spreadsheet", "Power": "Grouping from Top-down Microarchitecture Analysis Metrics sp= readsheet", + "Prefetches": "Grouping from Top-down Microarchitecture Analysis Metri= cs spreadsheet", "Ret": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", "Retire": "Grouping from Top-down Microarchitecture Analysis Metrics s= preadsheet", "SMT": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -75,6 +78,7 @@ "tma_bad_speculation_group": "Metrics contributing to tma_bad_speculat= ion category", "tma_branch_resteers_group": "Metrics contributing to tma_branch_reste= ers category", "tma_core_bound_group": "Metrics contributing to tma_core_bound catego= ry", + "tma_divider_group": "Metrics contributing to tma_divider category", "tma_dram_bound_group": "Metrics contributing to tma_dram_bound catego= ry", "tma_dtlb_load_group": "Metrics contributing to tma_dtlb_load category= ", "tma_dtlb_store_group": "Metrics contributing to tma_dtlb_store catego= ry", @@ -99,6 +103,7 @@ "tma_issueSmSt": "Metrics related by the issue $issueSmSt", "tma_issueSyncxn": "Metrics related by the issue $issueSyncxn", "tma_issueTLB": "Metrics related by the issue $issueTLB", + "tma_itlb_misses_group": "Metrics contributing to tma_itlb_misses cate= gory", "tma_l1_bound_group": "Metrics contributing to tma_l1_bound category", "tma_light_operations_group": "Metrics contributing to tma_light_opera= tions category", "tma_machine_clears_group": "Metrics contributing to tma_machine_clear= s category", diff --git a/tools/perf/pmu-events/arch/x86/jaketown/other.json b/tools/per= f/pmu-events/arch/x86/jaketown/other.json index 42692fa24b6c..970839a9c786 100644 --- a/tools/perf/pmu-events/arch/x86/jaketown/other.json +++ b/tools/perf/pmu-events/arch/x86/jaketown/other.json @@ -33,14 +33,6 @@ "SampleAfterValue": "2000003", "UMask": "0x2" }, - { - "BriefDescription": "Valid instructions written to IQ per cycle.", - "Counter": "0,1,2,3", - "EventCode": "0x17", - "EventName": "INSTS_WRITTEN_TO_IQ.INSTS", - "SampleAfterValue": "2000003", - "UMask": "0x1" - }, { "BriefDescription": "Cycles when L1 and L2 are locked due to UC or= split lock.", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 373972E336F for ; Wed, 26 Mar 2025 00:08:12 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947702; cv=none; b=Q3Mxib0YO8H5zRmW3FXQBQ88zULykiVVd89DY/kAfjpR3xiI6qLJvn4o7lTVXxP/a8Cms2+4Kovkljm5tEEtYYggtcKzB3aH4cUSGa99yQsQ938C4AiVOs7WlN63ZbPjabc5pBYvq5G/Nr9VvIrpiCLE0t10rB1swtWzNzZSfrM= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947702; c=relaxed/simple; bh=/Czwh93Cr1HYyvvd0zF2rmxOC+/u0jE7zz1lBJSkchQ=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=rS7wqeYMybVUNX4HolaCkoLlqZ3nVwtBXSEDcViADK3mc20jgZpp5jUNAEIGE0zqDS1gryAvGkf/N0p0RR4LyqDbKaIOHhFN/cTnS12mQTq2l93zSHwbPCXSpiDRY9fcM/iLO8C/VPZ4UBa0EcgsRyliCrs8JYLRcGYU6CtS7FE= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=1yI++PVV; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="1yI++PVV" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff7f9a0b9bso10773124a91.0 for ; Tue, 25 Mar 2025 17:08:12 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947691; x=1743552491; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=syocn0SO+jBBIkE/o0xIYKpUTldsfeCpYzBj4W1zZEk=; b=1yI++PVVEcJ7M/25CdW9OipX6lSFOzzN96o/nR3Jp98RNWpCmMaPP6sT9xKOcVQaOd UEHxplUrJgxx5sAJQobZjW/+91p2l4kHXcHLCyLbCVkj4UjAQqE312sxjN+8uDPWBbHv NkpWthv95FofSE+obg32fLZXsfadJSiWlRYpMZOrBbKg3Gy8pdY8qvOedNURGrvWd6V8 gbSuNo1vYOiVZK+ktbZXP287G8S7mbbXzuuD3QdO2y22xcQCdJGtS/lBUwuyPb2RescH TXAjGxSaPtodxLanqP3YbNWDwM4f7N5IxCjlJZHqACGolTgIDj1jipfHZd5+w/0x3v7V 7+/A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947691; x=1743552491; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=syocn0SO+jBBIkE/o0xIYKpUTldsfeCpYzBj4W1zZEk=; b=pEkARcixK/bA+QbEHc9FCvgBEZiPe+BpXgKUjN4nIu01e67DDhM9q8jzzpV6XL7sSz xonKCg0fzPawV40Pg1oW4II4IfZTlX9ea0BQeqYiHDtn7cLiO79K3lwpoa+d3bECZ7ZV PUX5CPqTp8zKtuOWU2ajm+Rumo70YmbpChTbbGrNQJi72erJpoO5sUlDhQJnnNp553Tf KKEpkzK0VDLUivN4Y6x9FlbGcBRW8Urt/98ZhvcSDCw2mHwRsow1nLSHYz3Nz24S3MDD LdFf8dLTjZP6JANKVqEgTpIm5aWNzBs8ousNLz9MfHN97lZBsQh1twOQGr6aBADD8xid rUUA== X-Forwarded-Encrypted: i=1; AJvYcCUeOInRCw/mSkM8HNaKd82/5t/Gmg8ZYqzYWXeL0zqBzsQ3IyQRsQGSzFo1o5ne+a+38vKUyPW52BGaFgU=@vger.kernel.org X-Gm-Message-State: AOJu0YxQnavU5pZjIpkThPuDamNZg1DG7/E1Aq4anSrM5+vwjA0h83pQ RUXP6C2QdrJ502wddKm+Fbv8oFvLf8HdXNNXaenvEo0q2RsQ3lDXMbDqfRZtB5ShcYryMNUz5Tk ifdGqEQ== X-Google-Smtp-Source: AGHT+IFaNJJ8bZtYDqkZh1Oe3pPAZNeuDbn7iIOFRbtOiQc0jacsIe18ukgl9oxJK9Xcg4IGkczFBVpE9xUA X-Received: from pgbdm10.prod.google.com ([2002:a05:6a02:d8a:b0:ad5:5004:836f]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a20:918a:b0:1ee:dcd3:80d7 with SMTP id adf61e73a8af0-1fe42c6ce38mr28028702637.0.1742947691485; Tue, 25 Mar 2025 17:08:11 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:19 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-22-irogers@google.com> Subject: [PATCH v3 21/35] perf vendor events: Update lunarlake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/lunarlake/cache.json | 122 ++++ .../arch/x86/lunarlake/lnl-metrics.json | 556 +++++++++--------- .../pmu-events/arch/x86/lunarlake/memory.json | 44 ++ .../pmu-events/arch/x86/lunarlake/other.json | 353 ----------- .../arch/x86/lunarlake/pipeline.json | 187 ++++++ 5 files changed, 630 insertions(+), 632 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/lunarlake/cache.json b/tools/pe= rf/pmu-events/arch/x86/lunarlake/cache.json index 15fb9921f4fc..4f783e7eb947 100644 --- a/tools/perf/pmu-events/arch/x86/lunarlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/lunarlake/cache.json @@ -417,6 +417,51 @@ "UMask": "0x22", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of LLC prefetches that were= throttled due to Dynamic Prefetch Throttling. The throttle requestor/sou= rce could be from the uncore/SOC or the Dead Block Predictor. Counts on a p= er core basis.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x29", + "EventName": "LLC_PREFETCHES_THROTTLED.DPT", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of LLC prefetches throttled= due to Demand Throttle Prefetcher. DTP Global Triggered with no Local Ove= rride. Counts on a per core basis.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x29", + "EventName": "LLC_PREFETCHES_THROTTLED.DTP", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of LLC prefetches not throt= tled by DTP due to local override. These prefetches may still be throttled= due to another throttler mechanism. Counts on a per core basis.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x29", + "EventName": "LLC_PREFETCHES_THROTTLED.DTP_OVERRIDE", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of LLC prefetches throttled= due to LLC hit rate in . Counts on a per core basis= .", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x29", + "EventName": "LLC_PREFETCHES_THROTTLED.HIT_RATE", + "SampleAfterValue": "1000003", + "UMask": "0x10", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of LLC prefetches throttled= due to exceeding the XQ threshold set by either XQ_THRESOLD_DTP or LLC_XQ_= THRESHOLD. Counts on a per core basis.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x29", + "EventName": "LLC_PREFETCHES_THROTTLED.XQ_THRESH", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_atom" + }, { "BriefDescription": "Cycles when L1D is locked", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1183,6 +1228,39 @@ "UMask": "0xf", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts writebacks of modified cachelines that= have any type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.COREWB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10008", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts writebacks of non-modified cachelines = that have any type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.COREWB_NONM.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x11000", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by mem side cache.", "Counter": "0,1,2,3,4,5,6,7", @@ -1194,6 +1272,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache where a snoop hit in another cores caches, data forwarding i= s required as the data is modified.", "Counter": "0,1,2,3", @@ -1227,6 +1327,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that were su= pplied by the L3 cache where a snoop hit in another cores caches, data forw= arding is required as the data is modified.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/lunarlake/lnl-metrics.json b/to= ols/perf/pmu-events/arch/x86/lunarlake/lnl-metrics.json index e748f839c4bd..f6c4ffad66b6 100644 --- a/tools/perf/pmu-events/arch/x86/lunarlake/lnl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/lunarlake/lnl-metrics.json @@ -89,7 +89,7 @@ "MetricExpr": "tma_core_bound", "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", - "MetricThreshold": "(tma_allocation_restriction >0.10) & ((tma_cor= e_bound >0.10) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -99,7 +99,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALL_P@ / (8 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound >0.10)", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%", @@ -111,7 +111,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.ALL_P@ / (8 * cpu_= atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation >0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%", @@ -122,7 +122,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_DETECT@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "MetricThreshold": "(tma_branch_detect >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -132,7 +132,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MISPREDICT@ / (8 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts >0.05) & ((tma_bad_spe= culation >0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -142,7 +142,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_RESTEER@ / (8 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", - "MetricThreshold": "(tma_branch_resteer >0.05) & ((tma_ifetch_late= ncy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -151,7 +151,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.CISC@ / (8 * cpu_atom@CPU= _CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc >0.05) & ((tma_ifetch_bandwidth >0.1= 0) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -160,7 +160,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS@ / (8 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound >0.10) & ((tma_backend_bound >= 0.10))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -170,7 +170,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.DECODE@ / (8 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", - "MetricThreshold": "(tma_decode >0.05) & ((tma_ifetch_bandwidth >0= .10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -179,7 +179,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.FASTNUKE@ / (8 * c= pu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", - "MetricThreshold": "(tma_fast_nuke >0.05) & ((tma_machine_clears >= 0.05) & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -189,7 +189,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ALL@ / (8 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound >0.20)", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -199,7 +199,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ICACHE@ / (8 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -208,7 +208,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH@ / (8 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", - "MetricThreshold": "(tma_ifetch_bandwidth >0.10) & ((tma_frontend_= bound >0.20))", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -218,7 +218,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_LATENCY@ / (8 * = cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", - "MetricThreshold": "(tma_ifetch_latency >0.15) & ((tma_frontend_bo= und >0.20))", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -578,7 +578,7 @@ "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "cpu_atom@CPU_CLK_UNHALTED.CORE_P@ / cpu_atom@CPU_CL= K_UNHALTED.CORE@", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9", "Unit": "cpu_atom" }, { @@ -617,7 +617,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ITLB_MISS@ / (8 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses >0.05) & ((tma_ifetch_latency= >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -626,7 +626,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS@ / = (8 * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears >0.05) & ((tma_bad_specula= tion >0.15))", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -636,7 +636,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.MEM_SCHEDULER@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", - "MetricThreshold": "(tma_mem_scheduler >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -645,7 +645,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER@ / (8 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", - "MetricThreshold": "(tma_non_mem_scheduler >0.10) & ((tma_resource= _bound >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -654,7 +654,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.NUKE@ / (8 * cpu_a= tom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", - "MetricThreshold": "(tma_nuke >0.05) & ((tma_machine_clears >0.05)= & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -663,7 +663,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.OTHER@ / (8 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", - "MetricThreshold": "(tma_other_fb >0.05) & ((tma_ifetch_bandwidth = >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -672,7 +672,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.PREDECODE@ / (8 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", - "MetricThreshold": "(tma_predecode >0.05) & ((tma_ifetch_bandwidth= >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -681,7 +681,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REGISTER@ / (8 * cpu_atom= @CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", - "MetricThreshold": "(tma_register >0.10) & ((tma_resource_bound >0= .20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -690,7 +690,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REORDER_BUFFER@ / (8 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", - "MetricThreshold": "(tma_reorder_buffer >0.10) & ((tma_resource_bo= und >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -699,7 +699,7 @@ "MetricExpr": "tma_backend_bound - tma_core_bound", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", - "MetricThreshold": "(tma_resource_bound >0.20) & ((tma_backend_bou= nd >0.10))", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -710,7 +710,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_RETIRING.ALL@ / (8 * cpu_atom@CPU_= CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "(tma_retiring >0.75)", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -720,12 +720,12 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.SERIALIZATION@ / (8 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", - "MetricThreshold": "(tma_serialization >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "cpu_core@UOPS_DISPATCHED.ALU@ / (6 * tma_info_threa= d_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -738,13 +738,13 @@ "MetricExpr": "78 * cpu_core@ASSISTS.ANY@ / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_threa= d_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -755,7 +755,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-be\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -767,18 +767,18 @@ { "BriefDescription": "This category represents fraction of slots wa= sted due to incorrect speculations", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-bad\\-spec / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-bad\\-spec@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20", @@ -795,16 +795,16 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_latency + tma_= split_loads + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_capacity + tma_l1_latency_dependency + tma_lock_latency + tma_split_= loads + tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", - "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_mem_b= andwidth, tma_sq_full", + "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_= system_dram_bw_use, tma_mem_bandwidth, tma_sq_full", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + t= ma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bo= und + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_latency_c= apacity / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + = tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb_full)= ) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l= 3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtl= b_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_l1_latency_cap= acity + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bou= nd * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_= fwd_blk + tma_l1_latency_dependency + tma_l1_latency_capacity + tma_lock_la= tency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_store_bou= nd / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_sto= re_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sharing + t= ma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_memory_boun= d * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_store_latency / (tma_store_latency + tm= a_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity= + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_sto= re_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_latency_c= apacity / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma= _l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtl= b_load + tma_fb_full + tma_l1_latency_capacity + tma_l1_latency_dependency = + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bou= nd * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_fb_ful= l + tma_l1_latency_capacity + tma_l1_latency_dependency + tma_lock_latency = + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bou= nd / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_sto= re_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_= split_stores + tma_store_latency + tma_streaming_stores)) + tma_memory_boun= d * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_f= alse_sharing + tma_split_stores + tma_store_latency + tma_streaming_stores)= ))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -813,16 +813,16 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy", + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: ", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_= clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_br= anch_mispredicts) / (tma_mispredicts_resteers + tma_clears_resteers + tma_u= nknown_branches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_reste= ers + tma_ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth *= tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_co= de", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cl= ears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_bran= ch_mispredicts) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unk= nown_branches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_miss= es + tma_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * t= ma_ms / (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code= ", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", @@ -830,7 +830,7 @@ }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_s= witches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_rest= eers * tma_other_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_r= esteers + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses= + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_= dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ls= d + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other= _nukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cp= u_core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / = (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_mic= rocode_sequencer / (tma_microcode_sequencer + tma_few_uops_instructions) * = (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_restee= rs * tma_other_mispredicts / tma_branch_mispredicts) / (tma_clears_resteers= + tma_mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers= + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_m= s_switches) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite = + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_bra= nch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_n= ukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cpu_= core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / (t= ma_divider + tma_ports_utilization + tma_serializing_operation) + tma_micro= code_sequencer / (tma_microcode_sequencer + tma_few_uops_instructions) * (t= ma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -839,7 +839,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_depende= ncy + tma_l1_latency_capacity + tma_lock_latency + tma_split_loads + tma_fb= _full)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_dtlb_store / (= tma_store_latency + tma_false_sharing + tma_split_stores + tma_streaming_st= ores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_fb_full + tma_l1_latency_capacity + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fw= d_blk)) + tma_memory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bo= und + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_dtlb_store / (= tma_dtlb_store + tma_false_sharing + tma_split_stores + tma_store_latency += tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -848,16 +848,16 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears", + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -870,11 +870,11 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls", + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls.", "Unit": "cpu_core" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (cpu_core@BR_INST_RETIRED.ALL= _BRANCHES@ + 2 * cpu_core@BR_INST_RETIRED.NEAR_CALL@ + cpu_core@INST_RETIRE= D.NOP@) / tma_info_thread_slots - tma_microcode_sequencer / (tma_microcode_= sequencer + tma_few_uops_instructions) * (tma_assists / tma_microcode_seque= ncer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -883,7 +883,7 @@ }, { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "cpu_core@topdown\\-br\\-mispredict@ / (cpu_core@top= down\\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-re= tiring@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -897,26 +897,26 @@ "MetricExpr": "cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_= thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C01@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C02@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -925,8 +925,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -935,99 +935,99 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@frontend_retired.l1i_miss@R / tma_info_thread_clks - tma_code_l2_miss)", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@FRONTEND_RETIRED.L1I_MISS@R / tma_info_thread_clks - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@front= end_retired.l2_miss@R / tma_info_thread_clks", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@FRONT= END_RETIRED.L2_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the (first level) ITLB was missed by instructions fetches, th= at later on hit in second-level TLB (STLB)", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@frontend_retired.itlb_miss@R / tma_info_thread_clks - tma_code_stlb_mis= s)", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@FRONTEND_RETIRED.ITLB_MISS@R / tma_info_thread_clks - tma_code_stlb_mis= s)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates the fraction of cycles = where the Second-level TLB (STLB) was missed by instruction fetches, perfor= ming a hardware page walk", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@fro= ntend_retired.stlb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@FRO= NTEND_RETIRED.STLB_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@ / (cpu_core@ITLB_MISSES.= WALK_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_4K@ / (cpu_core@ITLB_MISSES.WAL= K_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@br_misp_retired.cond_ntaken_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@BR_MISP_RETIRED.COND_NTAKEN_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_nt_mispredicts", - "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by backward-taken conditional branche= s", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@ * cpu= _core@br_misp_retired.cond_taken_bwd_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by backward-taken conditional branche= s.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@ * cpu= _core@BR_MISP_RETIRED.COND_TAKEN_BWD_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_bwd_mispredicts", - "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & tma_branc= h_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_bwd_mispredicts > 0.05 & (tma_bran= ch_mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by forward-taken conditional branches= ", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@ * cpu= _core@br_misp_retired.cond_taken_fwd_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by forward-taken conditional branches= .", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@ * cpu= _core@BR_MISP_RETIRED.COND_TAKEN_FWD_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_fwd_mispredicts", - "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & tma_branc= h_mispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_fwd_mispredicts > 0.05 & (tma_bran= ch_mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ *= cpu_core@mem_load_l3_hit_retired.xsnp_miss@R, cpu_core@MEM_LOAD_L3_HIT_RET= IRED.XSNP_MISS@ * (27 * tma_info_system_core_frequency) - 3 * tma_info_syst= em_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_miss@R else= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * (27 * tma_info_system_core_f= requency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM@ * cpu_core@mem_load_l3_hit_retired.xsnp_hitm@R, cpu= _core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * (28 * tma_info_system_core_frequ= ency) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit= _retired.xsnp_hitm@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * (28= * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequency)) *= (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS= @ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * min(= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R, 24 * tma_info_system_core_fre= quency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM@ * min(cpu_core@MEM_LO= AD_L3_HIT_RETIRED.XSNP_HITM@R, 25 * tma_info_system_core_frequency)) * (1 += cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2= ) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1038,17 +1038,17 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@= * cpu_core@mem_load_l3_hit_retired.xsnp_no_fwd@R, cpu_core@MEM_LOAD_L3_HIT= _RETIRED.XSNP_NO_FWD@ * (27 * tma_info_system_core_frequency) - 3 * tma_inf= o_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_no_fw= d@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * (27 * tma_info_sys= tem_core_frequency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@M= EM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * cpu_core@mem_load_l3_hit_retired.xsnp_fw= d@R, cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (28 * tma_info_system_cor= e_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load= _l3_hit_retired.xsnp_fwd@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ = * (28 * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequenc= y)) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1= _MISS@ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * mi= n(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R, 24 * tma_info_system_core= _frequency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * min(cpu_core@MEM= _LOAD_L3_HIT_RETIRED.XSNP_FWD@R, 25 * tma_info_system_core_frequency)) * (1= + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ /= 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1057,7 +1057,7 @@ "MetricExpr": "cpu_core@ARITH.DIV_ACTIVE@ / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1067,7 +1067,7 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.MEM@ / tma_info_thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1078,7 +1078,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1087,28 +1087,28 @@ "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / tma_in= fo_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * cp= u_core@mem_inst_retired.stlb_hit_loads@R, cpu_core@MEM_INST_RETIRED.STLB_HI= T_LOADS@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_loads@R else cpu_co= re@MEM_INST_RETIRED.STLB_HIT_LOADS@ * 7) / tma_info_thread_clks + tma_load_= stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * min(cpu= _core@MEM_INST_RETIRED.STLB_HIT_LOADS@R, 7) / tma_info_thread_clks + tma_lo= ad_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * c= pu_core@mem_inst_retired.stlb_hit_stores@R, cpu_core@MEM_INST_RETIRED.STLB_= HIT_STORES@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_stores@R else cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * 7) / tma_info_thread_clks + tma_= store_stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * min(cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@R, 7) / tma_info_thread_clks + tma_= store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1117,8 +1117,8 @@ "MetricExpr": "28 * tma_info_system_core_frequency * cpu_core@OCR.= DEMAND_RFO.L3_HIT.SNOOP_HITM@ / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1128,7 +1128,7 @@ "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1145,12 +1145,12 @@ }, { "BriefDescription": "This metric represents fraction of slots the = CPU was stalled due to Frontend latency issues", - "MetricExpr": "topdown\\-fetch\\-lat / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-fetch\\-lat@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend= _bound_group", "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1160,7 +1160,7 @@ "MetricGroup": "TopdownL3;tma_L3_group;tma_heavy_operations_group;= tma_issueD0", "MetricName": "tma_few_uops_instructions", "MetricThreshold": "tma_few_uops_instructions > 0.05 & tma_heavy_o= perations > 0.1", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring instructions that that are decoder into two or more= uops. This highly-correlates with the number of uops in such instructions", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring instructions that that are decoder into two or more= uops. This highly-correlates with the number of uops in such instructions.= Related metrics: tma_decoder0_alone", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1170,7 +1170,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1180,16 +1180,16 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "cpu_core@ARITH.FPDIV_ACTIVE@ / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1198,8 +1198,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.SCALAR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_ports_utili= zed_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1208,8 +1208,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, = tma_int_vector_256b, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1218,8 +1218,8 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b,= tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1228,41 +1228,41 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR\\,umask\\=3D0= x30@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b,= tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots wh= ere the processor's Frontend undersupplies its Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-fe\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-fe\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvFB;BvIO;Default;PGO;TmaL1;TopdownL1;tma_L1_group= ", "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound.", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.MACRO_= FUSED@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "cpu_core@topdown\\-heavy\\-ops@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1271,26 +1271,26 @@ "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / tma_info_thread_clks= ", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= ", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@br_misp_retired.indirect_call_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= .", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_call_mispredicts", - "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions", - "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@br_misp_retired.indirect_cost@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@br_misp_retired.indirect_call_cost@R) / tma_info_thread= _clks, 0)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions.= ", + "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_COST@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R) / tma_info_thread= _clks, 0)", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_jump_mispredicts", - "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1303,7 +1303,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_NTAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", @@ -1311,29 +1311,29 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional backward-taken branches (lower number means higher occurrence rate)= ", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional backward-taken branches (lower number means higher occurrence rate)= .", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN_BWD@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_bwd", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional forward-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional forward-taken branches (lower number means higher occurrence rate).= ", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN_FWD@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken_fwd", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.INDIRECT@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.RET@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -1357,7 +1357,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -1366,7 +1366,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -1375,10 +1375,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: ", "Unit": "cpu_core" }, { @@ -1444,12 +1445,12 @@ "MetricExpr": "(cpu_core@FP_ARITH_DISPATCHED.V0@ + cpu_core@FP_ARI= TH_DISPATCHED.V1@ + cpu_core@FP_ARITH_DISPATCHED.V2@ + cpu_core@FP_ARITH_DI= SPATCHED.V3@) / (4 * tma_info_thread_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)", + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n).", "Unit": "cpu_core" }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp", "Unit": "cpu_core" @@ -1464,15 +1465,15 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired DSB misses", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= frontend_retired.any_dsb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= FRONTEND_RETIRED.ANY_DSB_MISS@R / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat", "MetricName": "tma_info_frontend_dsb_switches_ret", "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05", @@ -1480,7 +1481,7 @@ }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D1@", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc", "Unit": "cpu_core" @@ -1530,7 +1531,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired operations that invoke th= e Microcode Sequencer", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@fron= tend_retired.ms_flows@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@FRON= TEND_RETIRED.MS_FLOWS@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq", "MetricName": "tma_info_frontend_ms_latency_ret", "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05", @@ -1545,21 +1546,21 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node", + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node.", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired branches who got branch a= ddress clears", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@frontend_retired.unknown_branch@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@FRONTEND_RETIRED.UNKNOWN_BRANCH@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat", "MetricName": "tma_info_frontend_unknown_branches_ret", "Unit": "cpu_core" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "cpu_core@BR_INST_RETIRED.ALL_BRANCHES@ / cpu_core@B= R_INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch", @@ -1579,7 +1580,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW", + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW.", "Unit": "cpu_core" }, { @@ -1588,7 +1589,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1597,7 +1598,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1606,7 +1607,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1615,7 +1616,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1678,7 +1679,7 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 8 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 17", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp", "Unit": "cpu_core" }, @@ -1803,7 +1804,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp", "Unit": "cpu_core" @@ -1861,7 +1862,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand loads", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@mem_inst_retired.stlb_miss_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@MEM_INST_RETIRED.STLB_MISS_LOADS@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_load_stlb_miss_ret > 0.05", @@ -1884,7 +1885,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand stores", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@mem_inst_retired.stlb_miss_stores@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@MEM_INST_RETIRED.STLB_MISS_STORES@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_store_stlb_miss_ret > 0.05= ", @@ -1923,20 +1924,20 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@ASSISTS.ANY@", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)", "Unit": "cpu_core" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire", "Unit": "cpu_core" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1", @@ -1981,23 +1982,22 @@ }, { "BriefDescription": "Instructions per Far Branch ( Far Branches ap= ply upon transition from application to operating system, handling interrup= ts, exceptions) [lower number means higher occurrence rate]", - "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / BR_INST_RETIRED.FAR_BR= ANCH:u", + "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.FAR_BRANCH@u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000", + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6", "Unit": "cpu_core" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@INS= T_RETIRED.ANY_P@k", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / cpu_core@CPU_CLK_UNHA= LTED.THREAD@", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@CPU= _CLK_UNHALTED.THREAD@", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05", @@ -2034,7 +2034,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD@", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks", @@ -2045,7 +2045,6 @@ "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { @@ -2053,7 +2052,7 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_ISSU= ED.ANY@", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage", + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage.", "Unit": "cpu_core" }, { @@ -2065,7 +2064,7 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "cpu_core@TOPDOWN.SLOTS@", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots", "Unit": "cpu_core" @@ -2083,15 +2082,15 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@BR_= INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 8 * 1.5", + "MetricThreshold": "tma_info_thread_uptb < 12", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2101,7 +2100,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2110,8 +2109,8 @@ "MetricExpr": "cpu_core@INT_VEC_RETIRED.128BIT@ / (tma_retiring * = tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2120,8 +2119,8 @@ "MetricExpr": "cpu_core@INT_VEC_RETIRED.256BIT@ / (tma_retiring * = tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_ports_utilized_= 2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2130,8 +2129,8 @@ "MetricExpr": "cpu_core@ICACHE_TAG.STALLS@ / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2140,17 +2139,17 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L1@ / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit Level 1 after missing Level 0 within the L1D= cache", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * cpu_cor= e@mem_load_retired.l1_hit_l1@R, cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * 9) i= f 0 < cpu_core@mem_load_retired.l1_hit_l1@R else cpu_core@MEM_LOAD_RETIRED.= L1_HIT_L1@ * 9) / tma_info_thread_clks", + "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit Level 1 after missing Level 0 within the L1D= cache.", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L1_HIT_L1@ * min(cpu_core= @MEM_LOAD_RETIRED.L1_HIT_L1@R, 9) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_capacity", - "MetricThreshold": "tma_l1_latency_capacity > 0.1 & tma_l1_bound >= 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_capacity > 0.1 & (tma_l1_bound = > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2159,8 +2158,8 @@ "MetricExpr": "4 * cpu_core@DEPENDENT_LOADS.ANY@ / tma_info_thread= _clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: DEPENDENT_LOADS.ANY", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_UOPS_RETIRED.L1_HIT_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2169,17 +2168,17 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L2@ / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles wit= h demand load accesses that hit the L2 cache under unloaded scenarios (poss= ibly L2 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * cpu_core@m= em_load_retired.l2_hit@R, cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info= _system_core_frequency)) if 0 < cpu_core@mem_load_retired.l2_hit@R else cpu= _core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info_system_core_frequency)) * (1= + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ /= 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L2_HIT@R, 3 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2189,18 +2188,18 @@ "MetricExpr": "cpu_core@MEMORY_STALLS.L3@ / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * cpu_core@m= em_load_retired.l3_hit@R, cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 * tma_inf= o_system_core_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_c= ore@mem_load_retired.l3_hit@R else cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 = * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequency) * (= 1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ = / 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L3_HIT@R, 9 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2209,19 +2208,19 @@ "MetricExpr": "cpu_core@DECODE.LCP@ / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2231,7 +2230,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_load_op_utilization", "MetricThreshold": "tma_load_op_utilization > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Load operations. Sample with: = UOPS_DISPATCHED.LOAD", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Load operations. Sample with: = UOPS_DISPATCHED.PORT_2_3", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2240,7 +2239,7 @@ "MetricExpr": "max(0, tma_dtlb_load - tma_load_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2249,43 +2248,43 @@ "MetricExpr": "cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@ / tma_info_t= hread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_1G@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_2M_4M@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_C= OMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_4K@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU spent handling cache misses due to lock operations", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@me= m_inst_retired.lock_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@ME= M_INST_RETIRED.LOCK_LOADS@R / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2296,7 +2295,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2307,17 +2306,17 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_sq_full", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2326,34 +2325,34 @@ "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD@) / tma_info_thread_clks - tm= a_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-mem\\-bound@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * cpu_core@MISC2_RETIRED.LFENCE@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * cpu_core@MEM_UOP_RETIRED.ANY= @ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -2376,14 +2375,14 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * cpu_= core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the MITE pipeline (the legacy deco= de pipeline)", - "MetricExpr": "(cpu_core@IDQ.MITE_UOPS\\,cmask\\=3D0x8\\,inv\\=3D0= x1@ / tma_info_thread_clks + cpu_core@IDQ.MITE_UOPS@ / (cpu_core@IDQ.DSB_UO= PS@ + cpu_core@IDQ.MITE_UOPS@) * (cpu_core@IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.= CORE@ - cpu_core@IDQ_BUBBLES.FETCH_LATENCY@)) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@IDQ.MITE_UOPS\\,cmask\\=3D0x8\\,inv\\=3D0= x1@ / 2 + cpu_core@IDQ.MITE_UOPS@ / (cpu_core@IDQ.DSB_UOPS@ + cpu_core@IDQ.= MITE_UOPS@) * (cpu_core@IDQ_BUBBLES.CYCLES_0_UOPS_DELIV.CORE@ - cpu_core@ID= Q_BUBBLES.FETCH_LATENCY@)) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL3;tma_L3_group;tma_fetch_b= andwidth_group", "MetricName": "tma_mite", "MetricThreshold": "tma_mite > 0.1 & tma_fetch_bandwidth > 0.2", @@ -2392,17 +2391,17 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", "MetricExpr": "cpu_core@IDQ.MS_CYCLES_ANY@ / tma_info_thread_clks", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", @@ -2415,7 +2414,7 @@ "MetricExpr": "3 * cpu_core@IDQ.MS_SWITCHES@ / tma_info_thread_clk= s", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2426,7 +2425,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2435,7 +2434,7 @@ "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.NOP@ /= (tma_retiring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2451,20 +2450,20 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - cpu_core@BR_MISP_= RETIRED.ALL_BRANCHES@ / (cpu_core@INT_MISC.CLEARS_COUNT@ - cpu_core@MACHINE= _CLEARS.COUNT@)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - cpu_core@MACHINE_CLEA= RS.MEMORY_ORDERING@ / cpu_core@MACHINE_CLEARS.COUNT@), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2474,7 +2473,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2483,8 +2482,8 @@ "MetricExpr": "((cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ + (cpu_c= ore@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVITY.2_3_P= ORTS_UTIL@)) / tma_info_thread_clks if cpu_core@ARITH.DIV_ACTIVE@ < cpu_cor= e@CYCLE_ACTIVITY.STALLS_TOTAL@ - cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ else= (cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVIT= Y.2_3_PORTS_UTIL@) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2493,8 +2492,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ / tma_info= _thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2503,7 +2502,7 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2514,8 +2513,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.2_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma= _int_vector_256b", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2525,24 +2524,24 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.CYCLES_GE_3@ / tma_info_thre= ad_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@br_mi= sp_retired.ret_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@BR_MI= SP_RETIRED.RET_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ret_mispredicts", - "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispre= dicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & (tma_branch_mispr= edicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-retiring@ / (cpu_core@topdown\\-= fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiring@= + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2556,8 +2555,8 @@ "MetricExpr": "(cpu_core@BE_STALLS.SCOREBOARD@ + cpu_core@CPU_CLK_= UNHALTED.C02@) / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: BE_STALLS.SCOREBOARD. Related metrics: tm= a_ms_switches", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related me= trics: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2566,8 +2565,8 @@ "MetricExpr": "tma_light_operations * cpu_core@INT_VEC_RETIRED.SHU= FFLES@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2577,28 +2576,28 @@ "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.PAUSE@ / tma_info_thread_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles hand= ling memory load split accesses - load that cross 64-byte cache line bounda= ry", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * cpu_c= ore@mem_inst_retired.split_loads@R, cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ = * tma_info_memory_load_miss_real_latency) if 0 < cpu_core@mem_inst_retired.= split_loads@R else cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * tma_info_memory= _load_miss_real_latency) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * min(cpu_co= re@MEM_INST_RETIRED.SPLIT_LOADS@R, tma_info_memory_load_miss_real_latency) = / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents rate of split store ac= cesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * cpu_= core@mem_inst_retired.split_stores@R, cpu_core@MEM_INST_RETIRED.SPLIT_STORE= S@) if 0 < cpu_core@mem_inst_retired.split_stores@R else cpu_core@MEM_INST_= RETIRED.SPLIT_STORES@) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * min(cpu_c= ore@MEM_INST_RETIRED.SPLIT_STORES@R, 1) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2607,8 +2606,8 @@ "MetricExpr": "(cpu_core@XQ.FULL@ + cpu_core@L1D_MISS.L2_STALLS@) = / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2617,8 +2616,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.BOUND_ON_STORES@ / tma_info_t= hread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2627,8 +2626,8 @@ "MetricExpr": "13 * cpu_core@LD_BLOCKS.STORE_FORWARD@ / tma_info_t= hread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2637,8 +2636,8 @@ "MetricExpr": "(cpu_core@MEM_STORE_RETIRED.L2_HIT@ * 10 * (1 - cpu= _core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.ALL_STORES@)= + (1 - cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.A= LL_STORES@) * min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFFCORE_REQUE= STS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO@)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2648,7 +2647,6 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_store_op_utilization", "MetricThreshold": "tma_store_op_utilization > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port for Store operations. Sample with:= UOPS_DISPATCHED.STD, UOPS_DISPATCHED.STA", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2657,7 +2655,7 @@ "MetricExpr": "max(0, tma_dtlb_store - tma_store_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2666,34 +2664,34 @@ "MetricExpr": "cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@ / tma_info_= thread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_1G@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_2M_4M@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_= core@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.W= ALK_COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_4K@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2702,7 +2700,7 @@ "MetricExpr": "9 * cpu_core@OCR.STREAMING_WR.ANY_RESPONSE@ / tma_i= nfo_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2712,7 +2710,7 @@ "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / tma_info= _thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2722,8 +2720,8 @@ "MetricExpr": "tma_retiring * cpu_core@UOPS_EXECUTED.X87@ / cpu_co= re@UOPS_EXECUTED.THREAD@", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%", "Unit": "cpu_core" } diff --git a/tools/perf/pmu-events/arch/x86/lunarlake/memory.json b/tools/p= erf/pmu-events/arch/x86/lunarlake/memory.json index 60daff922a89..589bd79fe069 100644 --- a/tools/perf/pmu-events/arch/x86/lunarlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/lunarlake/memory.json @@ -314,6 +314,17 @@ "UMask": "0x4", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1FBC000004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache and were su= pplied by the system memory (DRAM, MSC, or MMIO).", "Counter": "0,1,2,3,4,5,6,7", @@ -325,6 +336,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1FBC000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1E780000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the L3 cache and were supplied by the system memory (DRAM, MSC, or MM= IO).", "Counter": "0,1,2,3,4,5,6,7", @@ -347,6 +380,17 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that were su= pplied by DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1FBC000002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that were no= t supplied by the L3 cache and were supplied by the system memory (DRAM, MS= C, or MMIO).", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/lunarlake/other.json b/tools/pe= rf/pmu-events/arch/x86/lunarlake/other.json index 667707d4fe37..ad646db01f8c 100644 --- a/tools/perf/pmu-events/arch/x86/lunarlake/other.json +++ b/tools/perf/pmu-events/arch/x86/lunarlake/other.json @@ -18,15 +18,6 @@ "UMask": "0x8", "Unit": "cpu_core" }, - { - "BriefDescription": "Counts cycles where the pipeline is stalled d= ue to serializing operations.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa2", - "EventName": "BE_STALLS.SCOREBOARD", - "SampleAfterValue": "100003", - "UMask": "0x2", - "Unit": "cpu_core" - }, { "BriefDescription": "Counts the number of unhalted cycles a Core i= s blocked due to a lock In Progress issued by another core", "Counter": "0,1,2,3,4,5,6,7", @@ -65,15 +56,6 @@ "UMask": "0x8", "Unit": "cpu_atom" }, - { - "BriefDescription": "Count number of times a load is depending on = another load that had just write back its data or in previous or 2 cycles = back. This event supports in-direct dependency through a single uop.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x02", - "EventName": "DEPENDENT_LOADS.ANY", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, { "BriefDescription": "Counts the number of cycles the L2 Prefetcher= s are at throttle level 0", "Counter": "0,1,2,3,4,5,6,7", @@ -119,170 +101,6 @@ "UMask": "0x10", "Unit": "cpu_atom" }, - { - "BriefDescription": "Counts the number of uops executed on all Int= eger ports.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.ALL", - "SampleAfterValue": "1000003", - "UMask": "0xff", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on a load = port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.LD", - "PublicDescription": "Counts the number of uops executed on a load= port. This event counts for integer uops even if the destination is FP/ve= ctor", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 0.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.P0", - "SampleAfterValue": "1000003", - "UMask": "0x8", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 1.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.P1", - "SampleAfterValue": "1000003", - "UMask": "0x10", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 2.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.P2", - "SampleAfterValue": "1000003", - "UMask": "0x20", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 3.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.P3", - "SampleAfterValue": "1000003", - "UMask": "0x40", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on integer= port 0,1, 2, 3.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.PRIMARY", - "SampleAfterValue": "1000003", - "UMask": "0x78", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on a Store= address port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.STA", - "PublicDescription": "Counts the number of uops executed on a Stor= e address port. This event counts integer uops even if the data source is F= P/vector", - "SampleAfterValue": "1000003", - "UMask": "0x2", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of uops executed on an inte= ger store data and jump port.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb3", - "EventName": "INT_UOPS_EXECUTED.STD_JMP", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of LLC prefetches that were= throttled due to Dynamic Prefetch Throttling. The throttle requestor/sou= rce could be from the uncore/SOC or the Dead Block Predictor. Counts on a p= er core basis.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x29", - "EventName": "LLC_PREFETCHES_THROTTLED.DPT", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of LLC prefetches throttled= due to Demand Throttle Prefetcher. DTP Global Triggered with no Local Ove= rride. Counts on a per core basis.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x29", - "EventName": "LLC_PREFETCHES_THROTTLED.DTP", - "SampleAfterValue": "1000003", - "UMask": "0x2", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of LLC prefetches not throt= tled by DTP due to local override. These prefetches may still be throttled= due to another throttler mechanism. Counts on a per core basis.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x29", - "EventName": "LLC_PREFETCHES_THROTTLED.DTP_OVERRIDE", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of LLC prefetches throttled= due to LLC hit rate in . Counts on a per core basis= .", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x29", - "EventName": "LLC_PREFETCHES_THROTTLED.HIT_RATE", - "SampleAfterValue": "1000003", - "UMask": "0x10", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number of LLC prefetches throttled= due to exceeding the XQ threshold set by either XQ_THRESOLD_DTP or LLC_XQ_= THRESHOLD. Counts on a per core basis.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x29", - "EventName": "LLC_PREFETCHES_THROTTLED.XQ_THRESH", - "SampleAfterValue": "1000003", - "UMask": "0x8", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L1 cache (that is: no execution & load in flight = & no load missed L1 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L1", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L2 cache (that is: no execution & load in flight = & load missed L1 & no load missed L2 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L2", - "SampleAfterValue": "1000003", - "UMask": "0x2", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L3 cache (that is: no execution & load in flight = & load missed L1 & load missed L2 cache & no load missed L3 Cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.L3", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for Memory (that is: no execution & load in flight & = a load missed L3 cache)", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0x46", - "EventName": "MEMORY_STALLS.MEM", - "SampleAfterValue": "1000003", - "UMask": "0x8", - "Unit": "cpu_core" - }, { "BriefDescription": "Counts all requests that have any type of res= ponse.", "Counter": "0,1,2,3,4,5,6,7", @@ -294,127 +112,6 @@ "UMask": "0x1", "Unit": "cpu_atom" }, - { - "BriefDescription": "Counts writebacks of modified cachelines that= have any type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.COREWB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10008", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts writebacks of non-modified cachelines = that have any type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.COREWB_NONM.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x11000", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1FBC000004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1FBC000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1E780000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that were su= pplied by DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1FBC000002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, { "BriefDescription": "Counts full streaming stores (64 bytes, WCiLF= ) that have any type of response.", "Counter": "0,1,2,3,4,5,6,7", @@ -459,56 +156,6 @@ "UMask": "0x1", "Unit": "cpu_core" }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", - "Counter": "0,1,2,3,4,5,6,7,8,9", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "1000003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts the number issue slots not consumed d= ue to a color request for an FCW or MXCSR control register when all 4 colo= rs (copies) are already in use", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.COLOR_STALLS", - "SampleAfterValue": "1000003", - "UMask": "0x8", - "Unit": "cpu_atom" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3,4,5,6,7,8,9", diff --git a/tools/perf/pmu-events/arch/x86/lunarlake/pipeline.json b/tools= /perf/pmu-events/arch/x86/lunarlake/pipeline.json index f4ec7a884937..c6f41f5a5d62 100644 --- a/tools/perf/pmu-events/arch/x86/lunarlake/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/lunarlake/pipeline.json @@ -87,6 +87,15 @@ "UMask": "0x1f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts cycles where the pipeline is stalled d= ue to serializing operations.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa2", + "EventName": "BE_STALLS.SCOREBOARD", + "SampleAfterValue": "100003", + "UMask": "0x2", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts the total number of branch instruction= s retired for all branch types.", "Counter": "0,1,2,3,4,5,6,7", @@ -757,6 +766,15 @@ "UMask": "0x4", "Unit": "cpu_core" }, + { + "BriefDescription": "Count number of times a load is depending on = another load that had just write back its data or in previous or 2 cycles = back. This event supports in-direct dependency through a single uop.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x02", + "EventName": "DEPENDENT_LOADS.ANY", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, { "BriefDescription": "Cycles total of 1 uop is executed on all port= s and Reservation Station was not empty.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -992,6 +1010,89 @@ "UMask": "0x10", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of uops executed on all Int= eger ports.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.ALL", + "SampleAfterValue": "1000003", + "UMask": "0xff", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on a load = port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.LD", + "PublicDescription": "Counts the number of uops executed on a load= port. This event counts for integer uops even if the destination is FP/ve= ctor", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 0.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.P0", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 1.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.P1", + "SampleAfterValue": "1000003", + "UMask": "0x10", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 2.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.P2", + "SampleAfterValue": "1000003", + "UMask": "0x20", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 3.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.P3", + "SampleAfterValue": "1000003", + "UMask": "0x40", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on integer= port 0,1, 2, 3.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.PRIMARY", + "SampleAfterValue": "1000003", + "UMask": "0x78", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on a Store= address port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.STA", + "PublicDescription": "Counts the number of uops executed on a Stor= e address port. This event counts integer uops even if the data source is F= P/vector", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of uops executed on an inte= ger store data and jump port.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb3", + "EventName": "INT_UOPS_EXECUTED.STD_JMP", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, { "BriefDescription": "Number of vector integer instructions retired= of 128-bit vector-width.", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1267,6 +1368,42 @@ "UMask": "0x4", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L1 cache (that is: no execution & load in flight = & no load missed L1 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L1", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L2 cache (that is: no execution & load in flight = & load missed L1 & no load missed L2 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L2", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for L3 cache (that is: no execution & load in flight = & load missed L1 & load missed L2 cache & no load missed L3 Cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.L3", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts cycles where no execution is happening= due to loads waiting for Memory (that is: no execution & load in flight & = a load missed L3 cache)", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0x46", + "EventName": "MEMORY_STALLS.MEM", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_core" + }, { "BriefDescription": "LFENCE instructions retired", "Counter": "0,1,2,3,4,5,6,7,8,9", @@ -1393,6 +1530,56 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", + "Counter": "0,1,2,3,4,5,6,7,8,9", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number issue slots not consumed d= ue to a color request for an FCW or MXCSR control register when all 4 colo= rs (copies) are already in use", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.COLOR_STALLS", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts the number of issue slots not consumed= by the backend due to a micro-sequencer (MS) scoreboard, which stalls the = front-end from issuing from the UROM until a specified older uop retires.", "Counter": "0,1,2,3,4,5,6,7", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-oa1-f74.google.com (mail-oa1-f74.google.com [209.85.160.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 4D59F3C1F for ; Wed, 26 Mar 2025 00:08:21 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.160.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947711; cv=none; b=UITbgN2PdiCoToY2yOKvS8vBPSpIXNYvgj015FLQ2OXa2G973Fp/3YFHLrCOBsZmA/kiBLih2ld5lShzETmamxs1/DFLiBDdjtIQ651fBJenSs7ttrMCz/ENnzdzMsFi/I5qn9fDRyBNSYPBgnriugZwS9VzLSYlLCk0XHJGRUQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947711; c=relaxed/simple; bh=NQXbyveXRAYQ4oQXVMBUJlSuWwJTuoM3tiEVnhyE1kQ=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=XgaG9NBKQexGDZbjj3M4WIzC+OxnC+NnLHJFa0LL6dDVXg6dkIbDAQVj+TP+27dc5/UK8RLsxpcEVGbmtuEmY+I/4khgIO/YYVA23h5jrP8OIRS/5Our57MfNULib4Xcugn/VIH+mO80AfNY8JixJ7+BxwOjBJvgpjqG+TvOz5M= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=TXA5YEzM; arc=none smtp.client-ip=209.85.160.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="TXA5YEzM" Received: by mail-oa1-f74.google.com with SMTP id 586e51a60fabf-2c2339b26fcso4094364fac.3 for ; Tue, 25 Mar 2025 17:08:21 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947700; x=1743552500; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=2knIi3DAIvFucoWfXHgdLu99CrRQPtPV8YUIxeb7YIk=; b=TXA5YEzMdgXX5FhseYcS8JRoNqgSEkNXBNW4CIaWJpdGQZay/TWYLQ15xo6f6WqI2X GkpZkFJGIbq4mIpDlFGSCaJKKQt/NPs5kLwd4fVUuoT5UMQCuTu4Zz+Qzmqg8Ixva5BP KeDIoxicnd6XU5pOT/0fC6hjvVR/r6FV8+gT4wuOPovKHCTJcbtNmLfK5Xokh30Rv4jp dDlGJ8C9aSOXgKzz2jQdsAdOqogtMgg9h+PfWCyz4Zb/Smf2hAvJQzciSluiJUuDzKYo fpCKpB6HiLyZ9P8YN91C87hTiijz1+NPry7Kbis82bnfgvI+Qfc16GJqgf/TnCANWpuu SZ4g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947700; x=1743552500; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=2knIi3DAIvFucoWfXHgdLu99CrRQPtPV8YUIxeb7YIk=; b=kOIuA+Q/82Tyau1I89NPaJ7wY3QmTHt81OyLAaNKemLp6BrhxCH/WAiBqnpTUnapSJ 2O7Y4Kt2vVvjBzbrUOmKPpe3JdXpX15Pyo2/ArQsnjwPs2lJbV5EQcO3Ybn2YAT/bClF KqR05UVL5DuLnLEKKntzuoQBrVJUvOCo9LGAKE822kv/PsoGue9HYYh2L1uxREE1Jr5i 8xipl7iPuye/sbTruMyl5yC7ya90eGNs7LpFchy299M4bDFH9X7ow+BBaLEE8gHdY2/2 jPb5K6YvO9JKo01THDvbrlST53CXX3/mH+lG3ex5b/2YEiYa4O1A2i66KZMehIAh6e/t rKDg== X-Forwarded-Encrypted: i=1; AJvYcCWvaR58cHnnKUhyJhJwzGPKcS9jt8/8TXjnCLtfxjRZhG0NsgrnBNUi34dFeHfi3qYtWiS9m/ENlDFc9/A=@vger.kernel.org X-Gm-Message-State: AOJu0YxTLA/K/U+78MG/GiMlwkZdH70C1eU2WZQQJO7+Up6qeFYxzoPP GNjutyH3BAG7cNeNvBUwrNXWHuFk6ciaq9ODUF5jU5xIyWKMzMqixOqr8CcSASgH3wHRtZ5WFRA dDZg6DQ== X-Google-Smtp-Source: AGHT+IG1mUV1JYU/krzPyZi+L1PCOSUfR/dLxkzPHXJ5HPSJ74LUchqwee23fDCyHMeT51lKrICaYKEATnoY X-Received: from oabfs10.prod.google.com ([2002:a05:6870:f78a:b0:25c:cce1:889e]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6870:2a4b:b0:2c2:dfbb:2dbd with SMTP id 586e51a60fabf-2c7801bd206mr11317603fac.8.1742947700361; Tue, 25 Mar 2025 17:08:20 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:20 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-23-irogers@google.com> Subject: [PATCH v3 22/35] perf vendor events: Update meteorlake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.12 to v1.13. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- .../pmu-events/arch/x86/meteorlake/cache.json | 179 ++++++ .../arch/x86/meteorlake/memory.json | 44 ++ .../arch/x86/meteorlake/mtl-metrics.json | 549 +++++++++--------- .../pmu-events/arch/x86/meteorlake/other.json | 140 ----- .../arch/x86/meteorlake/pipeline.json | 44 +- .../arch/x86/meteorlake/uncore-memory.json | 18 + 7 files changed, 559 insertions(+), 417 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 579b4fbd65d6..0c16c9b840a5 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -23,7 +23,7 @@ GenuineIntel-6-3E,v24,ivytown,core GenuineIntel-6-2D,v24,jaketown,core GenuineIntel-6-(57|85),v16,knightslanding,core GenuineIntel-6-BD,v1.11,lunarlake,core -GenuineIntel-6-(AA|AC|B5),v1.12,meteorlake,core +GenuineIntel-6-(AA|AC|B5),v1.13,meteorlake,core GenuineIntel-6-1[AEF],v4,nehalemep,core GenuineIntel-6-2E,v4,nehalemex,core GenuineIntel-6-A7,v1.04,rocketlake,core diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/cache.json b/tools/p= erf/pmu-events/arch/x86/meteorlake/cache.json index ce351cd7caaf..7f455864b1a7 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/cache.json @@ -1,4 +1,14 @@ [ + { + "BriefDescription": "Counts the number of L1D cacheline (dirty) ev= ictions caused by load misses, stores, and prefetches.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x51", + "EventName": "DL1.DIRTY_EVICTION", + "PublicDescription": "Counts the number of L1D cacheline (dirty) e= victions caused by load misses, stores, and prefetches. Does not count evi= ctions or dirty writebacks caused by snoops. Does not count a replacement = unless a (dirty) line was written back.", + "SampleAfterValue": "200003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "L1D.HWPF_MISS", "Counter": "0,1,2,3", @@ -81,6 +91,56 @@ "UMask": "0x1f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Exclusive state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.E", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Exclusive state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Forward state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.F", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Forward state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x10", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Modified state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.M", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Modified state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x8", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of cache lines filled into = the L2 cache that are in Shared state", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x25", + "EventName": "L2_LINES_IN.S", + "PublicDescription": "Counts the number of cache lines filled into= the L2 cache that are in Shared state. Counts on a per core basis.", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = evicted due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.NON_SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= evicted due to an L2 cache fill. Increments on the core that brought the l= ine in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, { "BriefDescription": "Modified cache lines that are evicted by L2 c= ache when triggered by an L2 cache fill.", "Counter": "0,1,2,3", @@ -91,6 +151,16 @@ "UMask": "0x2", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 cache lines that are = silently dropped due to an L2 cache fill", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x26", + "EventName": "L2_LINES_OUT.SILENT", + "PublicDescription": "Counts the number of L2 cache lines that are= silently dropped due to an L2 cache fill. Increments on the core that bro= ught the line in originally.", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Non-modified cache lines that are silently dr= opped by L2 cache.", "Counter": "0,1,2,3", @@ -121,6 +191,15 @@ "UMask": "0xff", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that r= esulted in a Hit from a front door request only (does not include rejects o= r recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.HIT", + "SampleAfterValue": "200003", + "UMask": "0x2", + "Unit": "cpu_atom" + }, { "BriefDescription": "All requests that hit L2 cache. [This event i= s alias to L2_RQSTS.HIT]", "Counter": "0,1,2,3", @@ -131,6 +210,15 @@ "UMask": "0xdf", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of total L2 Cache Accesses = that resulted in a Miss from a front door request only (does not include re= jects or recycles), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.MISS", + "SampleAfterValue": "200003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Read requests with true-miss in L2 cache [Thi= s event is alias to L2_RQSTS.MISS]", "Counter": "0,1,2,3", @@ -141,6 +229,15 @@ "UMask": "0x3f", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts the number of L2 Cache Accesses that m= iss the L2 and get BBL reject short and long rejects (includes those count= ed in L2_reject_XQ.any), per core event", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x24", + "EventName": "L2_REQUEST.REJECTS", + "SampleAfterValue": "200003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, { "BriefDescription": "L2 code requests", "Counter": "0,1,2,3", @@ -398,6 +495,15 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts the number of cycles the core is stall= ed due to an instruction cache or TLB miss which missed in the L2 cache.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x35", + "EventName": "MEM_BOUND_STALLS_IFETCH.L2_MISS", + "SampleAfterValue": "1000003", + "UMask": "0x7e", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to an ICACHE or ITLB miss which hit in the LLC. If the= core has access to an L3 cache, an LLC hit refers to an L3 cache hit, othe= rwise it counts zeros.", "Counter": "0,1,2,3,4,5,6,7", @@ -435,6 +541,15 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts the number of cycles the core is stall= ed due to a demand load which missed in the L2 cache.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x34", + "EventName": "MEM_BOUND_STALLS_LOAD.L2_MISS", + "SampleAfterValue": "1000003", + "UMask": "0x7e", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled due to a demand load miss which hit in the LLC. If the cor= e has access to an L3 cache, an LLC hit refers to an L3 cache hit, otherwis= e it counts zeros.", "Counter": "0,1,2,3,4,5,6,7", @@ -453,6 +568,15 @@ "UMask": "0x78", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts the number of unhalted cycles when the= core is stalled to a store buffer full condition", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x34", + "EventName": "MEM_BOUND_STALLS_LOAD.SBFULL", + "SampleAfterValue": "1000003", + "UMask": "0x80", + "Unit": "cpu_atom" + }, { "BriefDescription": "Retired load instructions.", "Counter": "0,1,2,3", @@ -1054,6 +1178,17 @@ "UMask": "0x3", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", @@ -1098,6 +1233,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", @@ -1164,6 +1321,28 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/memory.json b/tools/= perf/pmu-events/arch/x86/meteorlake/memory.json index e4481fbc1e13..8f07575da9f0 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/memory.json @@ -294,6 +294,17 @@ "UMask": "0x4", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", @@ -305,6 +316,28 @@ "UMask": "0x1", "Unit": "cpu_atom" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_core" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", @@ -327,6 +360,17 @@ "UMask": "0x1", "Unit": "cpu_core" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1", + "Unit": "cpu_atom" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/mtl-metrics.json b/t= ools/perf/pmu-events/arch/x86/meteorlake/mtl-metrics.json index 20c52630127e..89111cfcf3ae 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/mtl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/mtl-metrics.json @@ -75,7 +75,7 @@ "MetricExpr": "tma_core_bound", "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", - "MetricThreshold": "(tma_allocation_restriction >0.10) & ((tma_cor= e_bound >0.10) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -85,7 +85,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALL_P@ / (6 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", - "MetricThreshold": "(tma_backend_bound >0.10)", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%", @@ -97,7 +97,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.ALL_P@ / (6 * cpu_= atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", - "MetricThreshold": "(tma_bad_speculation >0.15)", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%", @@ -108,7 +108,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_DETECT@ / (6 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "MetricThreshold": "(tma_branch_detect >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -118,7 +118,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MISPREDICT@ / (6 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", - "MetricThreshold": "(tma_branch_mispredicts >0.05) & ((tma_bad_spe= culation >0.15))", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -128,7 +128,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.BRANCH_RESTEER@ / (6 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", - "MetricThreshold": "(tma_branch_resteer >0.05) & ((tma_ifetch_late= ncy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -137,7 +137,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.CISC@ / (6 * cpu_atom@CPU= _CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", - "MetricThreshold": "(tma_cisc >0.05) & ((tma_ifetch_bandwidth >0.1= 0) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -146,7 +146,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS@ / (6 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", - "MetricThreshold": "(tma_core_bound >0.10) & ((tma_backend_bound >= 0.10))", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -156,7 +156,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.DECODE@ / (6 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", - "MetricThreshold": "(tma_decode >0.05) & ((tma_ifetch_bandwidth >0= .10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -165,7 +165,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.FASTNUKE@ / (6 * c= pu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", - "MetricThreshold": "(tma_fast_nuke >0.05) & ((tma_machine_clears >= 0.05) & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -175,7 +175,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ALL_P@ / (6 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", - "MetricThreshold": "(tma_frontend_bound >0.20)", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -185,7 +185,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ICACHE@ / (6 * cpu_atom@C= PU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "(tma_icache_misses >0.05) & ((tma_ifetch_laten= cy >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -194,7 +194,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH@ / (6 = * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", - "MetricThreshold": "(tma_ifetch_bandwidth >0.10) & ((tma_frontend_= bound >0.20))", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -204,7 +204,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.FRONTEND_LATENCY@ / (6 * = cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", - "MetricThreshold": "(tma_ifetch_latency >0.15) & ((tma_frontend_bo= und >0.20))", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -564,7 +564,7 @@ "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "cpu_atom@CPU_CLK_UNHALTED.CORE_P@ / cpu_atom@CPU_CL= K_UNHALTED.CORE@", "MetricName": "tma_info_system_mux", - "MetricThreshold": "((tma_info_system_mux > 1.1)|(tma_info_system_= mux < 0.9))", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9", "Unit": "cpu_atom" }, { @@ -603,7 +603,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.ITLB_MISS@ / (6 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "(tma_itlb_misses >0.05) & ((tma_ifetch_latency= >0.15) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -612,7 +612,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS@ / = (6 * cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", - "MetricThreshold": "(tma_machine_clears >0.05) & ((tma_bad_specula= tion >0.15))", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -622,7 +622,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.MEM_SCHEDULER@ / (6 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", - "MetricThreshold": "(tma_mem_scheduler >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -631,7 +631,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER@ / (6 *= cpu_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", - "MetricThreshold": "(tma_non_mem_scheduler >0.10) & ((tma_resource= _bound >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -640,7 +640,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BAD_SPECULATION.NUKE@ / (6 * cpu_a= tom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", - "MetricThreshold": "(tma_nuke >0.05) & ((tma_machine_clears >0.05)= & ((tma_bad_speculation >0.15)))", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -649,7 +649,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.OTHER@ / (6 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", - "MetricThreshold": "(tma_other_fb >0.05) & ((tma_ifetch_bandwidth = >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -658,7 +658,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_FE_BOUND.PREDECODE@ / (6 * cpu_ato= m@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", - "MetricThreshold": "(tma_predecode >0.05) & ((tma_ifetch_bandwidth= >0.10) & ((tma_frontend_bound >0.20)))", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -667,7 +667,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REGISTER@ / (6 * cpu_atom= @CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", - "MetricThreshold": "(tma_register >0.10) & ((tma_resource_bound >0= .20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -676,7 +676,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.REORDER_BUFFER@ / (6 * cp= u_atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", - "MetricThreshold": "(tma_reorder_buffer >0.10) & ((tma_resource_bo= und >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -685,7 +685,7 @@ "MetricExpr": "tma_backend_bound - tma_core_bound", "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", - "MetricThreshold": "(tma_resource_bound >0.20) & ((tma_backend_bou= nd >0.10))", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -696,7 +696,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_RETIRING.ALL_P@ / (6 * cpu_atom@CP= U_CLK_UNHALTED.CORE@)", "MetricGroup": "Default;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", - "MetricThreshold": "(tma_retiring >0.75)", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1;Default", "ScaleUnit": "100%", "Unit": "cpu_atom" @@ -706,7 +706,7 @@ "MetricExpr": "cpu_atom@TOPDOWN_BE_BOUND.SERIALIZATION@ / (6 * cpu= _atom@CPU_CLK_UNHALTED.CORE@)", "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", - "MetricThreshold": "(tma_serialization >0.10) & ((tma_resource_bou= nd >0.20) & ((tma_backend_bound >0.10)))", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%", "Unit": "cpu_atom" }, @@ -718,7 +718,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(cpu_core@UOPS_DISPATCHED.PORT_0@ + cpu_core@UOPS_D= ISPATCHED.PORT_1@ + cpu_core@UOPS_DISPATCHED.PORT_5_11@ + cpu_core@UOPS_DIS= PATCHED.PORT_6@) / (5 * tma_info_core_core_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -731,13 +731,13 @@ "MetricExpr": "78 * cpu_core@ASSISTS.ANY@ / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_threa= d_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -748,7 +748,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-be\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -765,13 +765,13 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20", @@ -788,16 +788,16 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", - "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_mem_b= andwidth, tma_sq_full", + "PublicDescription": "Total pipeline cost of external Memory- or C= ache-Bandwidth related bottlenecks. Related metrics: tma_fb_full, tma_info_= system_dram_bw_use, tma_mem_bandwidth, tma_sq_full", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_= bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_d= ram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_st= ore_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_load= s + tma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_= l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_s= tore_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sha= ring + tma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_mem= ory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound = + tma_dram_bound + tma_store_bound)) * (tma_store_latency / (tma_store_late= ncy + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtl= b_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependen= cy + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_= bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma= _l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_fb= _full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tm= a_store_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tm= a_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_= latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_s= tore_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharin= g + tma_split_stores + tma_store_latency + tma_streaming_stores)) + tma_mem= ory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store= + tma_false_sharing + tma_split_stores + tma_store_latency + tma_streaming= _stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -806,16 +806,16 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy", + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: ", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_= clears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_br= anch_mispredicts) / (tma_mispredicts_resteers + tma_clears_resteers + tma_u= nknown_branches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_reste= ers + tma_ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth *= tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_co= de", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - c= pu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D= 1@) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cl= ears_resteers + tma_mispredicts_resteers * tma_other_mispredicts / tma_bran= ch_mispredicts) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unk= nown_branches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_miss= es + tma_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * t= ma_ms / (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code= ", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20", @@ -823,7 +823,7 @@ }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_s= witches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_rest= eers * tma_other_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_r= esteers + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses= + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_= dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ls= d + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other= _nukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cp= u_core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / = (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_mic= rocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * = (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - cpu_core@INST_RETIRED.REP_ITERATION@ / = cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_restee= rs * tma_other_mispredicts / tma_branch_mispredicts) / (tma_clears_resteers= + tma_mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers= + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_m= s_switches) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite = + tma_ms)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_bra= nch_mispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_n= ukes / tma_other_nukes + tma_core_bound * (tma_serializing_operation + cpu_= core@RS.EMPTY_RESOURCE@ / tma_info_thread_clks * tma_ports_utilized_0) / (t= ma_divider + tma_ports_utilization + tma_serializing_operation) + tma_micro= code_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) * (t= ma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -832,7 +832,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_depende= ncy + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound= * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dra= m_bound + tma_store_bound)) * (tma_dtlb_store / (tma_store_latency + tma_fa= lse_sharing + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (= tma_dtlb_load / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + = tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound= * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l= 3_bound + tma_store_bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false= _sharing + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -841,16 +841,16 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears", + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache", "Unit": "cpu_core" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -863,11 +863,11 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls", + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls.", "Unit": "cpu_core" }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (cpu_core@BR_INST_RETIRED.ALL= _BRANCHES@ + 2 * cpu_core@BR_INST_RETIRED.NEAR_CALL@ + cpu_core@INST_RETIRE= D.NOP@) / tma_info_thread_slots - tma_microcode_sequencer / (tma_few_uops_i= nstructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_seque= ncer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -876,7 +876,7 @@ }, { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "cpu_core@topdown\\-br\\-mispredict@ / (cpu_core@top= down\\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-re= tiring@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TmaL2;TopdownL2;tma_L2_= group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -890,26 +890,26 @@ "MetricExpr": "cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_= thread_clks + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C01@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.C02@ / tma_info_thread_cl= ks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -918,7 +918,7 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRE= D.MS_FLOWS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -928,90 +928,90 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * cpu_core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@frontend_retired.l1i_miss@R / tma_info_thread_clks - tma_code_l2_miss)", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.L1I_MISS@ * cpu_co= re@FRONTEND_RETIRED.L1I_MISS@R / tma_info_thread_clks - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@front= end_retired.l2_miss@R / tma_info_thread_clks", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.L2_MISS@ * cpu_core@FRONT= END_RETIRED.L2_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the (first level) ITLB was missed by instructions fetches, th= at later on hit in second-level TLB (STLB)", - "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@frontend_retired.itlb_miss@R / tma_info_thread_clks - tma_code_stlb_mis= s)", + "MetricExpr": "max(0, cpu_core@FRONTEND_RETIRED.ITLB_MISS@ * cpu_c= ore@FRONTEND_RETIRED.ITLB_MISS@R / tma_info_thread_clks - tma_code_stlb_mis= s)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates the fraction of cycles = where the Second-level TLB (STLB) was missed by instruction fetches, perfor= ming a hardware page walk", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@fro= ntend_retired.stlb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.STLB_MISS@ * cpu_core@FRO= NTEND_RETIRED.STLB_MISS@R / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@ / (cpu_core@ITLB_MISSES.= WALK_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "cpu_core@ITLB_MISSES.WALK_ACTIVE@ / tma_info_thread= _clks * cpu_core@ITLB_MISSES.WALK_COMPLETED_4K@ / (cpu_core@ITLB_MISSES.WAL= K_COMPLETED_4K@ + cpu_core@ITLB_MISSES.WALK_COMPLETED_2M_4M@)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@br_misp_retired.cond_ntaken_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by non-taken conditional bran= ches.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_NTAKEN_COST@ * cpu_co= re@BR_MISP_RETIRED.COND_NTAKEN_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_nt_mispredicts", - "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_nt_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by taken conditional branches", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_COST@ * cpu_cor= e@br_misp_retired.cond_taken_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to misprediction by taken conditional branches.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.COND_TAKEN_COST@ * cpu_cor= e@BR_MISP_RETIRED.COND_TAKEN_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_cond_tk_mispredicts", - "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & tma_branch_mi= spredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_cond_tk_mispredicts > 0.05 & (tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ *= cpu_core@mem_load_l3_hit_retired.xsnp_miss@R, cpu_core@MEM_LOAD_L3_HIT_RET= IRED.XSNP_MISS@ * (27 * tma_info_system_core_frequency) - 3 * tma_info_syst= em_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_miss@R else= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * (27 * tma_info_system_core_f= requency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@MEM_LOAD_L3= _HIT_RETIRED.XSNP_FWD@ * cpu_core@mem_load_l3_hit_retired.xsnp_fwd@R, cpu_c= ore@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (28 * tma_info_system_core_frequenc= y) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit_re= tired.xsnp_fwd@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (28 * tm= a_info_system_core_frequency) - 3 * tma_info_system_core_frequency) * (cpu_= core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DEMAND_DATA_RD.L= 3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD@)= )) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_= MISS@ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@ * min(= cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS@R, 24 * tma_info_system_core_fre= quency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * min(cpu_core@MEM_LOA= D_L3_HIT_RETIRED.XSNP_FWD@R, 25 * tma_info_system_core_frequency) * (cpu_co= re@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DEMAND_DATA_RD.L3_= HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD@)))= * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MI= SS@ / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1022,26 +1022,26 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "((min(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@= * cpu_core@mem_load_l3_hit_retired.xsnp_no_fwd@R, cpu_core@MEM_LOAD_L3_HIT= _RETIRED.XSNP_NO_FWD@ * (27 * tma_info_system_core_frequency) - 3 * tma_inf= o_system_core_frequency) if 0 < cpu_core@mem_load_l3_hit_retired.xsnp_no_fw= d@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * (27 * tma_info_sys= tem_core_frequency) - 3 * tma_info_system_core_frequency) + (min(cpu_core@M= EM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * cpu_core@mem_load_l3_hit_retired.xsnp_fw= d@R, cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * (27 * tma_info_system_cor= e_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_core@mem_load= _l3_hit_retired.xsnp_fwd@R else cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ = * (27 * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequenc= y) * (1 - cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DE= MAND_DATA_RD.L3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_= HIT_WITH_FWD@))) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LO= AD_RETIRED.L1_MISS@ / 2) / tma_info_thread_clks", + "MetricExpr": "(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@ * mi= n(cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD@R, 24 * tma_info_system_core= _frequency) + cpu_core@MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD@ * min(cpu_core@MEM= _LOAD_L3_HIT_RETIRED.XSNP_FWD@R, 24 * tma_info_system_core_frequency) * (1 = - cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM@ / (cpu_core@OCR.DEMAND_DAT= A_RD.L3_HIT.SNOOP_HITM@ + cpu_core@OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH= _FWD@))) * (1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIR= ED.L1_MISS@ / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - = cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks = / 2", + "MetricExpr": "(cpu_core@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cp= u_core@INST_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1051,7 +1051,7 @@ "MetricExpr": "cpu_core@ARITH.DIV_ACTIVE@ / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIV_ACTIVE", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1061,7 +1061,7 @@ "MetricExpr": "cpu_core@MEMORY_ACTIVITY.STALLS_L3_MISS@ / tma_info= _thread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -1072,7 +1072,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1081,28 +1081,28 @@ "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / tma_in= fo_thread_clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * cp= u_core@mem_inst_retired.stlb_hit_loads@R, cpu_core@MEM_INST_RETIRED.STLB_HI= T_LOADS@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_loads@R else cpu_co= re@MEM_INST_RETIRED.STLB_HIT_LOADS@ * 7) / tma_info_thread_clks + tma_load_= stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_LOADS@ * min(cpu= _core@MEM_INST_RETIRED.STLB_HIT_LOADS@R, 7) / tma_info_thread_clks + tma_lo= ad_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * c= pu_core@mem_inst_retired.stlb_hit_stores@R, cpu_core@MEM_INST_RETIRED.STLB_= HIT_STORES@ * 7) if 0 < cpu_core@mem_inst_retired.stlb_hit_stores@R else cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * 7) / tma_info_thread_clks + tma_= store_stlb_miss", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_HIT_STORES@ * min(cp= u_core@MEM_INST_RETIRED.STLB_HIT_STORES@R, 7) / tma_info_thread_clks + tma_= store_stlb_miss", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1111,8 +1111,8 @@ "MetricExpr": "28 * tma_info_system_core_frequency * cpu_core@OCR.= DEMAND_RFO.L3_HIT.SNOOP_HITM@ / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1122,7 +1122,7 @@ "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_mem_bandwidth, tma_sq_full, tma_store_latency, tma_streaming_stores", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1133,18 +1133,18 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = CPU was stalled due to Frontend latency issues", - "MetricExpr": "topdown\\-fetch\\-lat / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) - cpu_core@I= NT_MISC.UOP_DROPPING@ / tma_info_thread_slots", + "MetricExpr": "cpu_core@topdown\\-fetch\\-lat@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) - cpu_core@INT_MISC.UOP_DROPPING@ / t= ma_info_thread_slots", "MetricGroup": "Frontend;TmaL2;TopdownL2;tma_L2_group;tma_frontend= _bound_group", "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1164,7 +1164,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1174,16 +1174,16 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "cpu_core@ARITH.FPDIV_ACTIVE@ / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1192,8 +1192,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.SCALAR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma= _port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1202,8 +1202,8 @@ "MetricExpr": "cpu_core@FP_ARITH_INST_RETIRED.VECTOR@ / (tma_retir= ing * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1212,8 +1212,8 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1222,41 +1222,41 @@ "MetricExpr": "(cpu_core@FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE@= + cpu_core@FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE@) / (tma_retiring * tm= a_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_int_vector_128b, tma_int_vector_256b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots wh= ere the processor's Frontend undersupplies its Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-fe\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) - cpu_core@IN= T_MISC.UOP_DROPPING@ / tma_info_thread_slots", + "MetricExpr": "cpu_core@topdown\\-fe\\-bound@ / (cpu_core@topdown\= \-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retirin= g@ + cpu_core@topdown\\-be\\-bound@) - cpu_core@INT_MISC.UOP_DROPPING@ / tm= a_info_thread_slots", "MetricGroup": "BvFB;BvIO;Default;PGO;TmaL1;TopdownL1;tma_L1_group= ", "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.MACRO_= FUSED@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "cpu_core@topdown\\-heavy\\-ops@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1265,26 +1265,26 @@ "MetricExpr": "cpu_core@ICACHE_DATA.STALLS@ / tma_info_thread_clks= ", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= ", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@br_misp_retired.indirect_call_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect CALL instructions= .", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_call_mispredicts", - "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_call_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions", - "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@br_misp_retired.indirect_cost@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@br_misp_retired.indirect_call_cost@R) / tma_info_thread= _clks, 0)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by indirect JMP instructions.= ", + "MetricExpr": "max((cpu_core@BR_MISP_RETIRED.INDIRECT_COST@ * cpu_= core@BR_MISP_RETIRED.INDIRECT_COST@R - cpu_core@BR_MISP_RETIRED.INDIRECT_CA= LL_COST@ * cpu_core@BR_MISP_RETIRED.INDIRECT_CALL_COST@R) / tma_info_thread= _clks, 0)", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ind_jump_mispredicts", - "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & tma_branch_m= ispredicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ind_jump_mispredicts > 0.05 & (tma_branch_= mispredicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -1297,7 +1297,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_NTAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", @@ -1305,7 +1305,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.COND_TAKEN@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", @@ -1313,15 +1313,15 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.INDIRECT@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000", + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3", "Unit": "cpu_core" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_MISP_RETIR= ED.RET@", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -1353,7 +1353,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -1362,7 +1362,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -1371,10 +1371,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: ", "Unit": "cpu_core" }, { @@ -1445,12 +1446,12 @@ "MetricExpr": "(cpu_core@FP_ARITH_DISPATCHED.PORT_0@ + cpu_core@FP= _ARITH_DISPATCHED.PORT_1@ + cpu_core@FP_ARITH_DISPATCHED.PORT_5@) / (2 * tm= a_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)", + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n).", "Unit": "cpu_core" }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_EXEC= UTED.THREAD\\,cmask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp", "Unit": "cpu_core" @@ -1465,15 +1466,15 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "cpu_core@DSB2MITE_SWITCHES.PENALTY_CYCLES@ / cpu_co= re@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired DSB misses", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= frontend_retired.any_dsb_miss@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.ANY_DSB_MISS@ * cpu_core@= FRONTEND_RETIRED.ANY_DSB_MISS@R / tma_info_thread_clks", "MetricGroup": "DSBmiss;Fed;FetchLat", "MetricName": "tma_info_frontend_dsb_switches_ret", "MetricThreshold": "tma_info_frontend_dsb_switches_ret > 0.05", @@ -1481,7 +1482,7 @@ }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@UOPS_ISSUED.ANY@ / cpu_core@UOPS_ISSUED.AN= Y\\,cmask\\=3D1@", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc", "Unit": "cpu_core" @@ -1531,7 +1532,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired operations that invoke th= e Microcode Sequencer", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@fron= tend_retired.ms_flows@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.MS_FLOWS@ * cpu_core@FRON= TEND_RETIRED.MS_FLOWS@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat;MicroSeq", "MetricName": "tma_info_frontend_ms_latency_ret", "MetricThreshold": "tma_info_frontend_ms_latency_ret > 0.05", @@ -1546,21 +1547,21 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / cpu_core= @INT_MISC.UNKNOWN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node", + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node.", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to retired branches who got branch a= ddress clears", - "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@frontend_retired.unknown_branch@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@FRONTEND_RETIRED.UNKNOWN_BRANCH@ * cpu_cor= e@FRONTEND_RETIRED.UNKNOWN_BRANCH@R / tma_info_thread_clks", "MetricGroup": "Fed;FetchLat", "MetricName": "tma_info_frontend_unknown_branches_ret", "Unit": "cpu_core" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "cpu_core@BR_INST_RETIRED.ALL_BRANCHES@ / cpu_core@B= R_INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch", @@ -1580,7 +1581,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW", + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW.", "Unit": "cpu_core" }, { @@ -1589,7 +1590,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1598,7 +1599,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1607,7 +1608,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1616,7 +1617,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting", + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting.", "Unit": "cpu_core" }, { @@ -1679,7 +1680,7 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 13", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp", "Unit": "cpu_core" }, @@ -1825,7 +1826,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_R= D@ / cpu_core@OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp", "Unit": "cpu_core" @@ -1883,7 +1884,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand loads", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@mem_inst_retired.stlb_miss_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_LOADS@ * cpu_co= re@MEM_INST_RETIRED.STLB_MISS_LOADS@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_load_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_load_stlb_miss_ret > 0.05", @@ -1906,7 +1907,7 @@ }, { "BriefDescription": "This metric represents fraction of cycles the= CPU retirement was stalled likely due to STLB misses by demand stores", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@mem_inst_retired.stlb_miss_stores@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.STLB_MISS_STORES@ * cpu_c= ore@MEM_INST_RETIRED.STLB_MISS_STORES@R / tma_info_thread_clks", "MetricGroup": "Mem;MemoryTLB", "MetricName": "tma_info_memory_tlb_store_stlb_miss_ret", "MetricThreshold": "tma_info_memory_tlb_store_stlb_miss_ret > 0.05= ", @@ -1921,7 +1922,7 @@ }, { "BriefDescription": "", - "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / (cpu_core@UOPS_EXE= CUTED.CORE_CYCLES_GE_1@ / 2 if #SMT_on else cpu_core@UOPS_EXECUTED.THREAD\\= ,cmask\\=3D0x1@)", + "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / (cpu_core@UOPS_EXE= CUTED.CORE_CYCLES_GE_1@ / 2 if #SMT_on else cpu_core@UOPS_EXECUTED.THREAD\\= ,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute", "Unit": "cpu_core" @@ -1952,20 +1953,20 @@ "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@ASSISTS.ANY@", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)", "Unit": "cpu_core" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire", "Unit": "cpu_core" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D0x1@", + "MetricExpr": "cpu_core@INST_RETIRED.REP_ITERATION@ / cpu_core@UOP= S_RETIRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1", @@ -2018,23 +2019,22 @@ }, { "BriefDescription": "Instructions per Far Branch ( Far Branches ap= ply upon transition from application to operating system, handling interrup= ts, exceptions) [lower number means higher occurrence rate]", - "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / BR_INST_RETIRED.FAR_BR= ANCH:u", + "MetricExpr": "cpu_core@INST_RETIRED.ANY@ / cpu_core@BR_INST_RETIR= ED.FAR_BRANCH@u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000", + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6", "Unit": "cpu_core" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@INS= T_RETIRED.ANY_P@k", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / cpu_core@CPU_CLK_UNHA= LTED.THREAD@", + "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD_P@k / cpu_core@CPU= _CLK_UNHALTED.THREAD@", "MetricGroup": "OS", "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05", @@ -2042,7 +2042,7 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_ARB_DAT_OCCUPANCY.RD / UNC_ARB_DAT_OCCUPANCY.RD= @thresh\\=3D0x1@", + "MetricExpr": "UNC_ARB_DAT_OCCUPANCY.RD / UNC_ARB_DAT_OCCUPANCY.RD= @cmask\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches", @@ -2093,7 +2093,7 @@ "Unit": "cpu_core" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.THREAD@", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks", @@ -2104,7 +2104,6 @@ "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr", "Unit": "cpu_core" }, { @@ -2112,7 +2111,7 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.THREAD@ / cpu_core@UOPS_ISSU= ED.ANY@", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage", + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage.", "Unit": "cpu_core" }, { @@ -2124,14 +2123,14 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "cpu_core@TOPDOWN.SLOTS@", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots", "Unit": "cpu_core" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (cpu_core@TOPDOWN.SLOTS@ /= 2) if #SMT_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization", "Unit": "cpu_core" @@ -2149,15 +2148,15 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu_core@BR_= INST_RETIRED.NEAR_TAKEN@", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5", + "MetricThreshold": "tma_info_thread_uptb < 9", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2167,7 +2166,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2176,8 +2175,8 @@ "MetricExpr": "(cpu_core@INT_VEC_RETIRED.ADD_128@ + cpu_core@INT_V= EC_RETIRED.VNNI_128@) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_int_vector_256b, tma_port_0, tma_port_1,= tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2186,8 +2185,8 @@ "MetricExpr": "(cpu_core@INT_VEC_RETIRED.ADD_256@ + cpu_core@INT_V= EC_RETIRED.MUL_256@ + cpu_core@INT_VEC_RETIRED.VNNI_256@) / (tma_retiring *= tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_port_0, tma_por= t_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2196,8 +2195,8 @@ "MetricExpr": "cpu_core@ICACHE_TAG.STALLS@ / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2206,7 +2205,7 @@ "MetricExpr": "max((cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ - cpu_co= re@MEMORY_ACTIVITY.STALLS_L1D_MISS@) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2216,7 +2215,7 @@ "MetricExpr": "min(2 * (cpu_core@MEM_INST_RETIRED.ALL_LOADS@ - cpu= _core@MEM_LOAD_RETIRED.FB_HIT@ - cpu_core@MEM_LOAD_RETIRED.L1_MISS@) * 20 /= 100, max(cpu_core@CYCLE_ACTIVITY.CYCLES_MEM_ANY@ - cpu_core@MEMORY_ACTIVIT= Y.CYCLES_L1D_MISS@, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2226,17 +2225,17 @@ "MetricExpr": "(cpu_core@MEMORY_ACTIVITY.STALLS_L1D_MISS@ - cpu_co= re@MEMORY_ACTIVITY.STALLS_L2_MISS@) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles wit= h demand load accesses that hit the L2 cache under unloaded scenarios (poss= ibly L2 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * cpu_core@m= em_load_retired.l2_hit@R, cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info= _system_core_frequency)) if 0 < cpu_core@mem_load_retired.l2_hit@R else cpu= _core@MEM_LOAD_RETIRED.L2_HIT@ * (3 * tma_info_system_core_frequency)) * (1= + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ /= 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L2_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L2_HIT@R, 3 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2246,18 +2245,18 @@ "MetricExpr": "(cpu_core@MEMORY_ACTIVITY.STALLS_L2_MISS@ - cpu_cor= e@MEMORY_ACTIVITY.STALLS_L3_MISS@) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(min(cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * cpu_core@m= em_load_retired.l3_hit@R, cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 * tma_inf= o_system_core_frequency) - 3 * tma_info_system_core_frequency) if 0 < cpu_c= ore@mem_load_retired.l3_hit@R else cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * (12 = * tma_info_system_core_frequency) - 3 * tma_info_system_core_frequency) * (= 1 + cpu_core@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ = / 2) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_LOAD_RETIRED.L3_HIT@ * min(cpu_core@ME= M_LOAD_RETIRED.L3_HIT@R, 9 * tma_info_system_core_frequency) * (1 + cpu_cor= e@MEM_LOAD_RETIRED.FB_HIT@ / cpu_core@MEM_LOAD_RETIRED.L1_MISS@ / 2) / tma_= info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2266,19 +2265,19 @@ "MetricExpr": "cpu_core@DECODE.LCP@ / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2297,7 +2296,7 @@ "MetricExpr": "max(0, tma_dtlb_load - tma_load_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2306,43 +2305,43 @@ "MetricExpr": "cpu_core@DTLB_LOAD_MISSES.WALK_ACTIVE@ / tma_info_t= hread_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_1G@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_2M_4M@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_C= OMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * cpu_core@DTLB_LOAD_MISSES.WALK= _COMPLETED_4K@ / (cpu_core@DTLB_LOAD_MISSES.WALK_COMPLETED_4K@ + cpu_core@D= TLB_LOAD_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_LOAD_MISSES.WALK_COMP= LETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU spent handling cache misses due to lock operations", - "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@me= m_inst_retired.lock_loads@R / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ * cpu_core@ME= M_INST_RETIRED.LOCK_LOADS@R / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2353,7 +2352,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2364,17 +2363,17 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_sq_full", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2383,34 +2382,34 @@ "MetricExpr": "min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFF= CORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DATA_RD@) / tma_info_thread_clks - tm= a_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-mem\\-bound@ / (cpu_core@topdown= \\-fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiri= ng@ + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "Backend;TmaL2;TopdownL2;tma_L2_group;tma_backend_b= ound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * cpu_core@MISC2_RETIRED.LFENCE@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * cpu_core@MEM_UOP_RETIRED.ANY= @ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -2433,7 +2432,7 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * cpu_= core@INT_MISC.CLEAR_RESTEER_CYCLES@ / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2449,18 +2448,18 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * cpu_core@ASSISTS.SSE_AVX_MIX@ / tma_info_thre= ad_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "max(cpu_core@IDQ.MS_CYCLES_ANY@, cpu_core@UOPS_RETI= RED.MS\\,cmask\\=3D0x1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISS= UED.ANY@)) / tma_info_core_core_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "max(cpu_core@IDQ.MS_CYCLES_ANY@, cpu_core@UOPS_RETI= RED.MS\\,cmask\\=3D1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUE= D.ANY@)) / tma_info_core_core_clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -2469,10 +2468,10 @@ }, { "BriefDescription": "This metric estimates the fraction of cycles = when the CPU was stalled due to switches of uop delivery to the Microcode S= equencer (MS)", - "MetricExpr": "3 * cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D0x1\\,edge= \\=3D0x1@ / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUED.ANY@) / tm= a_info_thread_clks", + "MetricExpr": "3 * cpu_core@UOPS_RETIRED.MS\\,cmask\\=3D1\\,edge@ = / (cpu_core@UOPS_RETIRED.SLOTS@ / cpu_core@UOPS_ISSUED.ANY@) / tma_info_thr= ead_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tm= a_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_mac= hine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_o= peration", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2483,7 +2482,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2492,7 +2491,7 @@ "MetricExpr": "tma_light_operations * cpu_core@INST_RETIRED.NOP@ /= (tma_retiring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2508,20 +2507,20 @@ "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - cpu_core@BR_MISP_= RETIRED.ALL_BRANCHES@ / (cpu_core@INT_MISC.CLEARS_COUNT@ - cpu_core@MACHINE= _CLEARS.COUNT@)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - cpu_core@MACHINE_CLEA= RS.MEMORY_ORDERING@ / cpu_core@MACHINE_CLEARS.COUNT@), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2531,7 +2530,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2541,7 +2540,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_12= 8b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_5, tma_po= rt_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2551,7 +2550,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vector_256b, tma_port_= 0, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2561,7 +2560,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128= b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= t_5, tma_ports_utilized_2", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2570,8 +2569,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (cp= u_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTIVITY.2_= 3_PORTS_UTIL@)) / tma_info_thread_clks if cpu_core@ARITH.DIV_ACTIVE@ < cpu_= core@CYCLE_ACTIVITY.STALLS_TOTAL@ - cpu_core@EXE_ACTIVITY.BOUND_ON_LOADS@ e= lse (cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ + tma_retiring * cpu_core@EXE_ACTI= VITY.2_3_PORTS_UTIL@) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2580,8 +2579,8 @@ "MetricExpr": "max(cpu_core@EXE_ACTIVITY.EXE_BOUND_0_PORTS@ - cpu_= core@RESOURCE_STALLS.SCOREBOARD@, 0) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2590,7 +2589,7 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.1_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2601,8 +2600,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.2_PORTS_UTIL@ / tma_info_thre= ad_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma= _int_vector_256b, tma_port_0, tma_port_1, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2612,24 +2611,24 @@ "MetricExpr": "cpu_core@UOPS_EXECUTED.CYCLES_GE_3@ / tma_info_thre= ad_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s", - "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@br_mi= sp_retired.ret_cost@R / tma_info_thread_clks", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to retired misprediction by (indirect) RET instruction= s.", + "MetricExpr": "cpu_core@BR_MISP_RETIRED.RET_COST@ * cpu_core@BR_MI= SP_RETIRED.RET_COST@R / tma_info_thread_clks", "MetricGroup": "BrMispredicts;TopdownL3;tma_L3_group;tma_branch_mi= spredicts_group", "MetricName": "tma_ret_mispredicts", - "MetricThreshold": "tma_ret_mispredicts > 0.05 & tma_branch_mispre= dicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_ret_mispredicts > 0.05 & (tma_branch_mispr= edicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "cpu_core@topdown\\-retiring@ / (cpu_core@topdown\\-= fe\\-bound@ + cpu_core@topdown\\-bad\\-spec@ + cpu_core@topdown\\-retiring@= + cpu_core@topdown\\-be\\-bound@) + 0 * tma_info_thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2643,7 +2642,7 @@ "MetricExpr": "cpu_core@RESOURCE_STALLS.SCOREBOARD@ / tma_info_thr= ead_clks + tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2653,8 +2652,8 @@ "MetricExpr": "tma_light_operations * cpu_core@INT_VEC_RETIRED.SHU= FFLES@ / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2664,28 +2663,28 @@ "MetricExpr": "cpu_core@CPU_CLK_UNHALTED.PAUSE@ / tma_info_thread_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric estimates fraction of cycles hand= ling memory load split accesses - load that cross 64-byte cache line bounda= ry", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * cpu_c= ore@mem_inst_retired.split_loads@R, cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ = * tma_info_memory_load_miss_real_latency) if 0 < cpu_core@mem_inst_retired.= split_loads@R else cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * tma_info_memory= _load_miss_real_latency) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_LOADS@ * min(cpu_co= re@MEM_INST_RETIRED.SPLIT_LOADS@R, tma_info_memory_load_miss_real_latency) = / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, { "BriefDescription": "This metric represents rate of split store ac= cesses", - "MetricExpr": "(min(cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * cpu_= core@mem_inst_retired.split_stores@R, cpu_core@MEM_INST_RETIRED.SPLIT_STORE= S@) if 0 < cpu_core@mem_inst_retired.split_stores@R else cpu_core@MEM_INST_= RETIRED.SPLIT_STORES@) / tma_info_thread_clks", + "MetricExpr": "cpu_core@MEM_INST_RETIRED.SPLIT_STORES@ * min(cpu_c= ore@MEM_INST_RETIRED.SPLIT_STORES@R, 1) / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2694,8 +2693,8 @@ "MetricExpr": "(cpu_core@XQ.FULL_CYCLES@ + cpu_core@L1D_PEND_MISS.= L2_STALLS@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_mem_bandwidth", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2704,8 +2703,8 @@ "MetricExpr": "cpu_core@EXE_ACTIVITY.BOUND_ON_STORES@ / tma_info_t= hread_clks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2714,8 +2713,8 @@ "MetricExpr": "13 * cpu_core@LD_BLOCKS.STORE_FORWARD@ / tma_info_t= hread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2724,8 +2723,8 @@ "MetricExpr": "(cpu_core@MEM_STORE_RETIRED.L2_HIT@ * 10 * (1 - cpu= _core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.ALL_STORES@)= + (1 - cpu_core@MEM_INST_RETIRED.LOCK_LOADS@ / cpu_core@MEM_INST_RETIRED.A= LL_STORES@) * min(cpu_core@CPU_CLK_UNHALTED.THREAD@, cpu_core@OFFCORE_REQUE= STS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO@)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2744,7 +2743,7 @@ "MetricExpr": "max(0, tma_dtlb_store - tma_store_stlb_miss)", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2753,34 +2752,34 @@ "MetricExpr": "cpu_core@DTLB_STORE_MISSES.WALK_ACTIVE@ / tma_info_= core_core_clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_1G@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_2M_4M@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_= core@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.W= ALK_COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * cpu_core@DTLB_STORE_MISSES.WA= LK_COMPLETED_4K@ / (cpu_core@DTLB_STORE_MISSES.WALK_COMPLETED_4K@ + cpu_cor= e@DTLB_STORE_MISSES.WALK_COMPLETED_2M_4M@ + cpu_core@DTLB_STORE_MISSES.WALK= _COMPLETED_1G@)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%", "Unit": "cpu_core" }, @@ -2789,7 +2788,7 @@ "MetricExpr": "9 * cpu_core@OCR.STREAMING_WR.ANY_RESPONSE@ / tma_i= nfo_thread_clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2799,7 +2798,7 @@ "MetricExpr": "cpu_core@INT_MISC.UNKNOWN_BRANCH_CYCLES@ / tma_info= _thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%", "Unit": "cpu_core" @@ -2809,8 +2808,8 @@ "MetricExpr": "tma_retiring * cpu_core@UOPS_EXECUTED.X87@ / cpu_co= re@UOPS_EXECUTED.THREAD@", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%", "Unit": "cpu_core" } diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/other.json b/tools/p= erf/pmu-events/arch/x86/meteorlake/other.json index 46a21776a4e9..4d64bedb3e8c 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/other.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/other.json @@ -28,105 +28,6 @@ "UMask": "0x1", "Unit": "cpu_atom" }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, - { - "BriefDescription": "Counts demand read for ownership (RFO) reques= ts and software prefetches for exclusive ownership (PREFETCHW) that have an= y type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1", - "Unit": "cpu_atom" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3,4,5,6,7", @@ -171,47 +72,6 @@ "UMask": "0x1", "Unit": "cpu_core" }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1", - "Unit": "cpu_core" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4", - "Unit": "cpu_atom" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/pipeline.json b/tool= s/perf/pmu-events/arch/x86/meteorlake/pipeline.json index 265f6c5a0248..e98a0324a6a6 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/pipeline.json @@ -1133,8 +1133,9 @@ "Unit": "cpu_atom" }, { - "BriefDescription": "Counts the number of machine clears that flus= h the pipeline and restart the machine with the use of microcode due to SMC= , MEMORY_ORDERING, FP_ASSISTS, PAGE_FAULT, DISAMBIGUATION, and FPC_VIRTUAL_= TRAP.", + "BriefDescription": "This event is deprecated.", "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", "EventCode": "0xc3", "EventName": "MACHINE_CLEARS.SLOW", "SampleAfterValue": "20003", @@ -1208,6 +1209,47 @@ "UMask": "0x2", "Unit": "cpu_core" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Cycles when RS was empty and a resource alloc= ation stall is asserted", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1", + "Unit": "cpu_core" + }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4", + "Unit": "cpu_atom" + }, { "BriefDescription": "This event counts a subset of the Topdown Slo= ts event that were not consumed by the back-end pipeline due to lack of bac= k-end resources, as a result of memory subsystem delays, execution units li= mitations, or other conditions.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/meteorlake/uncore-memory.json b= /tools/perf/pmu-events/arch/x86/meteorlake/uncore-memory.json index 783a4f7fd05b..ceb8839f0767 100644 --- a/tools/perf/pmu-events/arch/x86/meteorlake/uncore-memory.json +++ b/tools/perf/pmu-events/arch/x86/meteorlake/uncore-memory.json @@ -99,6 +99,24 @@ "PerPkg": "1", "Unit": "iMC" }, + { + "BriefDescription": "Any Rank at Hot state", + "Counter": "0,1,2,3,4", + "EventCode": "0x19", + "EventName": "UNC_M_DRAM_THERMAL_HOT", + "Experimental": "1", + "PerPkg": "1", + "Unit": "iMC" + }, + { + "BriefDescription": "Any Rank at Warm state", + "Counter": "0,1,2,3,4", + "EventCode": "0x1A", + "EventName": "UNC_M_DRAM_THERMAL_WARM", + "Experimental": "1", + "PerPkg": "1", + "Unit": "iMC" + }, { "BriefDescription": "PRE command sent to DRAM due to page table id= le timer expiration", "Counter": "0,1,2,3,4", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 1B6189475 for ; Wed, 26 Mar 2025 00:08:29 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947712; cv=none; b=Q2oK0hKxBiLjE6wUlVEG6YC0SJY0yo+2zxn5ZrZCuKHUZVbDfCrk4DDoJPkrx8kuWC4cKOfngb1zvbYrrWYfDPAHxgvNoXeqCZhqrWdWlELd5ORyA5THUC1BeKF+ARwYxFcG251lJxYjvSaJTyiMPUpjn4w60Ban2RkwBDTksJw= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947712; c=relaxed/simple; bh=8ECLI9MH8G6GS+TQuT+0Q+DWep0ey0sdDAegVmCJBw4=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=ePYuK75oBvg8VtllvQ50mj7931k8fKsS5CtLqDDGBmJO/YwtrE/Q+OXOSHnztHv7Bs7JtrRB84xSxDjVf32nLI//9RmXJVmb3rVDqvDpfqPRTpgTabt6Pk7C5mxc8lyPDrpbOGlsd29ThncRSMjulrZngM0jUpwz8P5Vl7Nz5L8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=sZfEbQwg; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="sZfEbQwg" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff7cf599beso12032708a91.0 for ; Tue, 25 Mar 2025 17:08:29 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947709; x=1743552509; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=NTidQqUKCPDZIA1M8BUtNh4exn5q3fZouzrUCwjO3OM=; b=sZfEbQwg9/rWEAMk55mHbrreSb5j/KD7k96TLJNcRkBz+LG0PpSYXNOB6OplnpDFYp pz0Ubecn6E52f6FiQBc4Vrh8MkkJ3b1ElJ5z8niSAasWGAPibO9OoN6cQGZyakXVLLi2 bQaoAGbJRgV6obj5/a5rTuuLQP0ri8mz7WSvWp9uk89fAV106g/7vDdwE5Oe45KL7npT Fl4HUwv8thTq581UFLiePepgPBvl31hYwy88wnwnGaMDhBRZlz3N8DxFUQw9FshmNMnn 3dXaMSujf3VmTN+IvDr3dlCQDbUk0XGBoVco2XSbCHjA/jZ4hD7KsLUAPKef8GLbOKqu PYkg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947709; x=1743552509; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=NTidQqUKCPDZIA1M8BUtNh4exn5q3fZouzrUCwjO3OM=; b=LVfB7fnFM+G0BF+HT6MhZhRfuEIRxMMygfVsXzr+FbSY6PKU7RjWV454uTdJPgrU5i jmdwHza3MCntncm9DQ5PMn61VpMVNu0O9afuSpfl8T4K6Be69rxaD/VIN9Kq5SZ097bL vjg+Jk4HzWec2wtuUmLjAFLp4J/9MHqISy/YznQVUlcSpl1sKcJFt/z/TnYWLFqhlY+o h06JGRMd8fpcIQbN6CpWXoTNg0gowSAmpn+GCD0wookQQLa4SgHL87c6ZkZura89CrCj ovKe41InRkJQ6n12w0lFIjM+rOW2bywYbgzUWpwmNdcsqBAO9ZAS+2Hhhwcd2dlNNETi W7UQ== X-Forwarded-Encrypted: i=1; AJvYcCVOXDBR4SE/Vi24CnJBg4Ddm+30SBgfNu5torS8h3Z08Hm3ooMWeUlztUJbFRBo34u90ZqTy6LEmEt+Iek=@vger.kernel.org X-Gm-Message-State: AOJu0Yze8aF85MMi5yDYLoeRUCEUQUj8nUL9h3x5OOtwl4J9X5HPuRe/ GNV8A1Gl3q/0hjgih+afngeCgC2IXdtcU44RnX6riRiJyF1qOj/UhmvFlkerDTbIEdEC1e1mMHV MLNKcPw== X-Google-Smtp-Source: AGHT+IHju4jV+R4MSc4FY1DMNx0mZaSamn1X9LuuzGH/pNZeR4xRlmoICP+RJ9tRjsBFyq2CAwPb/Hw0bheo X-Received: from pjbsj16.prod.google.com ([2002:a17:90b:2d90:b0:301:a339:b558]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:3d88:b0:2ff:556f:be8 with SMTP id 98e67ed59e1d1-3030fea91b0mr35591159a91.16.1742947709359; Tue, 25 Mar 2025 17:08:29 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:21 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-24-irogers@google.com> Subject: [PATCH v3 23/35] perf vendor events: Update nehalemep events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and virtual memory. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/nehalemep/cache.json | 32 +++++++++++++++ .../pmu-events/arch/x86/nehalemep/other.json | 40 ------------------- .../arch/x86/nehalemep/virtual-memory.json | 8 ++++ 3 files changed, 40 insertions(+), 40 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/nehalemep/cache.json b/tools/pe= rf/pmu-events/arch/x86/nehalemep/cache.json index b90026df2ce7..c9d154f1d09a 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemep/cache.json +++ b/tools/perf/pmu-events/arch/x86/nehalemep/cache.json @@ -239,6 +239,38 @@ "SampleAfterValue": "100000", "UMask": "0x2" }, + { + "BriefDescription": "L1I instruction fetch stall cycles", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.CYCLES_STALLED", + "SampleAfterValue": "2000000", + "UMask": "0x4" + }, + { + "BriefDescription": "L1I instruction fetch hits", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.HITS", + "SampleAfterValue": "2000000", + "UMask": "0x1" + }, + { + "BriefDescription": "L1I instruction fetch misses", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.MISSES", + "SampleAfterValue": "2000000", + "UMask": "0x2" + }, + { + "BriefDescription": "L1I Instruction fetches", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.READS", + "SampleAfterValue": "2000000", + "UMask": "0x3" + }, { "BriefDescription": "All L2 data requests", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/nehalemep/other.json b/tools/pe= rf/pmu-events/arch/x86/nehalemep/other.json index f6887b234b0e..5fe5ca778e9f 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemep/other.json +++ b/tools/perf/pmu-events/arch/x86/nehalemep/other.json @@ -15,46 +15,6 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, - { - "BriefDescription": "L1I instruction fetch stall cycles", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.CYCLES_STALLED", - "SampleAfterValue": "2000000", - "UMask": "0x4" - }, - { - "BriefDescription": "L1I instruction fetch hits", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.HITS", - "SampleAfterValue": "2000000", - "UMask": "0x1" - }, - { - "BriefDescription": "L1I instruction fetch misses", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.MISSES", - "SampleAfterValue": "2000000", - "UMask": "0x2" - }, - { - "BriefDescription": "L1I Instruction fetches", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.READS", - "SampleAfterValue": "2000000", - "UMask": "0x3" - }, - { - "BriefDescription": "Large ITLB hit", - "Counter": "0,1,2,3", - "EventCode": "0x82", - "EventName": "LARGE_ITLB.HIT", - "SampleAfterValue": "200000", - "UMask": "0x1" - }, { "BriefDescription": "All loads dispatched", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/nehalemep/virtual-memory.json b= /tools/perf/pmu-events/arch/x86/nehalemep/virtual-memory.json index e88c0802e679..accd263cfbfd 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemep/virtual-memory.json +++ b/tools/perf/pmu-events/arch/x86/nehalemep/virtual-memory.json @@ -88,6 +88,14 @@ "SampleAfterValue": "200000", "UMask": "0x20" }, + { + "BriefDescription": "Large ITLB hit", + "Counter": "0,1,2,3", + "EventCode": "0x82", + "EventName": "LARGE_ITLB.HIT", + "SampleAfterValue": "200000", + "UMask": "0x1" + }, { "BriefDescription": "Retired loads that miss the DTLB (Precise Eve= nt)", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B44D1E567 for ; Wed, 26 Mar 2025 00:08:38 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947721; cv=none; b=RyH+PkBxdxq5JhU5vdNTmXStwWevzEosKWOULv4hrqZDGMCn9ITmSdMmhxEcTKXi/PJ2CeUaDk3gLIgHIWMH+nd0YENxwg5SvLTv9H25406e+VmAROAd35ZsojAsTKeuQFsS7ugV7/YaGz9vHk7RyJMEFbq+Eg90HdPVcKsx+7c= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947721; c=relaxed/simple; bh=Zy0jRafJxRLJGt4ZHFy+/ySVLU1klhzlJJANwxxpiE0=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=V3Jx88B7YBp3+mPHcEuYJ97uykxZmOuzgbSWOwqWnigC3BJ5Eye+6NBaiaNhIE1zDM6LE4oMA5Xqq1TBMGhu143pdc9TNXSW+2qS4Pl0fuE1WUDzQgXNHuT1fHLO98vJ0/Vzg0jT5gINPIXxfyuSNmDfwkrAVY4NF3d9jQmGqT4= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=aWmbI1/0; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="aWmbI1/0" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2264e5b2b7cso93788435ad.2 for ; Tue, 25 Mar 2025 17:08:38 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947718; x=1743552518; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=kWTSLGRSQc6JnxhtpbR0m8qTat7TUVK+43G8X5ABzRg=; b=aWmbI1/0IE1wfvQQ5unrKbb0v9yhs003nX1kfI47f8m7NXO++zbDfqCnNQYWfEGmPp lIuyjBb0yQG4luHl3i5T+l6yEERnZhQtlkyRKxL132MMOzfa1o2EnTWzJBSFi/tZFnK6 OUumF7kq1xSsXexjYB4HsOvIdQmP273TLlnJs0hDpjIS0d50EY7OtFmG5eURGPEgsQ7a I5mPNM825JBfLMkUyjgVgTmHzyFwRr9mvsb5DazRLuMz8rCQ7tnDL0ruRjj8yIednLPq htgTzpjlaX+eC4eMGjWVA5eQSwh9gvW9jiCwBqOTM/sLGHYd5XrjN9SIk9pWfGqOmzEC j9ag== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947718; x=1743552518; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=kWTSLGRSQc6JnxhtpbR0m8qTat7TUVK+43G8X5ABzRg=; b=tr2X724u9WID12piF6oxZgABl+7i8/vb8IHuHARGZDkkFAluFjfgnDfo0gT64bTOHG UGD1rhhgbjIP+IRT0zanB90NJ1GBMcWqoRqNTvnekFFwEBXZnqRI0sYyLSLPqZqQsGKM x7S5U+B3RznYZU7SRBxHYGBZh78mhq5NUbQrLXv3BhxruFo253VgXUk1bYkX0/+8HxWF VPmgugrMPXVIKoZdP8fuaFnfv/TKhtIIIcTW76HE/6B+Hth54dcnt/pHEV8l+AGaa/dg m/gW8+kIKKBPIIkzioaa08c0YP7J2tvQsqmjWXtDJgrUrHR8WYhKCI8c44L96k1fg+Nj mxTg== X-Forwarded-Encrypted: i=1; AJvYcCXX+1a3heV+1jcdz95IZgYEA0EMCsc9/Y1V/XfS/4JA4klHIyT5N8f/uCqHvPnelX52Bv2V667k89Hm57I=@vger.kernel.org X-Gm-Message-State: AOJu0Yzv3i3yMT7DLJa5MAqTrRL3oC+i6Ej04crNIGldsPCjTFzbnsHV yxpVgo3uNsvJOnEcIn6JYLFelo0eQXMIzKxW9u6N7gx88y1ZOmZyyn5IyFXfXGiXuul8HzDh9lT Fn/iMug== X-Google-Smtp-Source: AGHT+IHloYEajhMtWC8hG1puIlY5v+wg4/tY5K4aV5d1Aea9Uw6SEBzc2wmss5k2k7hZqE9FrWd4Z/8m0D2a X-Received: from plpm4.prod.google.com ([2002:a17:903:3de4:b0:223:55e1:4789]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:902:f642:b0:211:e812:3948 with SMTP id d9443c01a7336-22780ba2e00mr289913145ad.0.1742947718236; Tue, 25 Mar 2025 17:08:38 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:22 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-25-irogers@google.com> Subject: [PATCH v3 24/35] perf vendor events: Update nehalemex events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and virtual memory. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/nehalemex/cache.json | 32 +++++++++++++++ .../pmu-events/arch/x86/nehalemex/other.json | 40 ------------------- .../arch/x86/nehalemex/virtual-memory.json | 8 ++++ 3 files changed, 40 insertions(+), 40 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/nehalemex/cache.json b/tools/pe= rf/pmu-events/arch/x86/nehalemex/cache.json index 2c0ea6f8c4e0..b6c6b22a3188 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemex/cache.json +++ b/tools/perf/pmu-events/arch/x86/nehalemex/cache.json @@ -239,6 +239,38 @@ "SampleAfterValue": "100000", "UMask": "0x2" }, + { + "BriefDescription": "L1I instruction fetch stall cycles", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.CYCLES_STALLED", + "SampleAfterValue": "2000000", + "UMask": "0x4" + }, + { + "BriefDescription": "L1I instruction fetch hits", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.HITS", + "SampleAfterValue": "2000000", + "UMask": "0x1" + }, + { + "BriefDescription": "L1I instruction fetch misses", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.MISSES", + "SampleAfterValue": "2000000", + "UMask": "0x2" + }, + { + "BriefDescription": "L1I Instruction fetches", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.READS", + "SampleAfterValue": "2000000", + "UMask": "0x3" + }, { "BriefDescription": "All L2 data requests", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/nehalemex/other.json b/tools/pe= rf/pmu-events/arch/x86/nehalemex/other.json index f6887b234b0e..5fe5ca778e9f 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemex/other.json +++ b/tools/perf/pmu-events/arch/x86/nehalemex/other.json @@ -15,46 +15,6 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, - { - "BriefDescription": "L1I instruction fetch stall cycles", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.CYCLES_STALLED", - "SampleAfterValue": "2000000", - "UMask": "0x4" - }, - { - "BriefDescription": "L1I instruction fetch hits", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.HITS", - "SampleAfterValue": "2000000", - "UMask": "0x1" - }, - { - "BriefDescription": "L1I instruction fetch misses", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.MISSES", - "SampleAfterValue": "2000000", - "UMask": "0x2" - }, - { - "BriefDescription": "L1I Instruction fetches", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.READS", - "SampleAfterValue": "2000000", - "UMask": "0x3" - }, - { - "BriefDescription": "Large ITLB hit", - "Counter": "0,1,2,3", - "EventCode": "0x82", - "EventName": "LARGE_ITLB.HIT", - "SampleAfterValue": "200000", - "UMask": "0x1" - }, { "BriefDescription": "All loads dispatched", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/nehalemex/virtual-memory.json b= /tools/perf/pmu-events/arch/x86/nehalemex/virtual-memory.json index e88c0802e679..accd263cfbfd 100644 --- a/tools/perf/pmu-events/arch/x86/nehalemex/virtual-memory.json +++ b/tools/perf/pmu-events/arch/x86/nehalemex/virtual-memory.json @@ -88,6 +88,14 @@ "SampleAfterValue": "200000", "UMask": "0x20" }, + { + "BriefDescription": "Large ITLB hit", + "Counter": "0,1,2,3", + "EventCode": "0x82", + "EventName": "LARGE_ITLB.HIT", + "SampleAfterValue": "200000", + "UMask": "0x1" + }, { "BriefDescription": "Retired loads that miss the DTLB (Precise Eve= nt)", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pl1-f202.google.com (mail-pl1-f202.google.com [209.85.214.202]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B5D1B16F288 for ; Wed, 26 Mar 2025 00:08:49 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947736; cv=none; b=Y06Ofp86UewglRT5HzEwJdWLRi69jvX7HAUckN3AnZ9VkqEKS7X4eNq+DMe8wMCn6A6peyGN6IWwfxSm8uvcm2MVf+3gFvflSQel9A/tncmwqEwvl3gWXO0Fbt4O/3TbRViGkuWPIW68rn3K4zaQErR2gsxbR4W2Vju8ydOU9pg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947736; c=relaxed/simple; bh=vUsqIw7sq2MnoeK2keexpdKbPhRbEDYXRLGzphWtfnE=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=SL0tW7vJXK/yff62lKq4ybKxOIwSsDEidR+rL9JEiOGEJM0LIemSV7vSaqPYR3gcFAZw2VtVkC2k+ursBHOPO4JIqyZJtUbzTyC1yNaJbz6tWIO9OObZDEqS75O3d4gYRX6N4ob9BMN4a/dNYtZMlRIAaYUqrJDwc2VrUjCAGzM= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=uraV0O+c; arc=none smtp.client-ip=209.85.214.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="uraV0O+c" Received: by mail-pl1-f202.google.com with SMTP id d9443c01a7336-2240a960f9cso101960775ad.0 for ; Tue, 25 Mar 2025 17:08:49 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947729; x=1743552529; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=2IP+cCqCUJLVI6CQa7WeJ28k5iF1L55KDmdpsl9mkcQ=; b=uraV0O+c0xTM5Ei7TVywLitq7uQzimG9bbck4mWMUibjx7hujRviC/SHAqSw/qllJt 2CFSOc+XRzR4tKeP8T4K174AiQYtP5jMgKRjBfXpunQC+hJ8usX4ggU4dZB5Bf0vHbq9 PQILd++2Zc6y0adzAig1CVHiaaw1/5tfA7JsciEsvIm9KQb0Fqu+OWz58odxIgbASxTH uzAVA8lAP8SbAp71Q9j/wwdCWAWoxXxupOqJRUDLwdkbVqmF0Xa3IYRf5BKU9SV93u6o QzLC39dSGWHiuEJwrk4Ji70e0uS1bBglmAM/Se8KNlNGD2KwVkbf+ATd/s2syZbsNTKQ zbKg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947729; x=1743552529; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=2IP+cCqCUJLVI6CQa7WeJ28k5iF1L55KDmdpsl9mkcQ=; b=qDpRcOIlKzoer1XQKs158x6xEg+LybgWa240VbrHRMra01xxYZ/2j14YSU6+y1tRyc 6tIMD/JuU6e4mqFO7WSgdoin/gmHen0CxAAYMrqJLAKoTay3rqbHLkOGI5O37ujZ011x gkrtYT8SAelWy/q1qHMMfk/550ow2BM9v2hqjmO/CfAeajmrNaNV56C5Ev1eu21ZgzxR frTU3gtv1I29/+xVWcQ95AG09tIKHVE+Q7T+AqxV1MJipI5VQVtncLMoS/TVux44Gc6c RWGgz2R3dNFIX/C8IwjDCsgQIVOu8pTQWmYbD/lcwHqSeavkrh3qN03HvYQCt4AD8W+Z nZZg== X-Forwarded-Encrypted: i=1; AJvYcCWlgZrp/apY09XqXplvbwKgXtZnzvXoQRdOzMHFDk1ojEAL7y8qxci42IFjvetHOf49v2OpDdXtfiHKHe8=@vger.kernel.org X-Gm-Message-State: AOJu0YwvD5dZTSwN63KqrVuzSoBu6G97XMnBDhH2lyEjYvPVE2yGrjPR DpIdsxKIfoGFUUOaBmfLyxT4tybb1j+feP9fBwgm34HF9jjEvR9YAl8kTzWy7yW9BCkjRIjXhDg cC9Qvmg== X-Google-Smtp-Source: AGHT+IEwKiw2aQvOxrzo/qDaUidDDGPR0T19+fx7rcCB9a4o1qRjZI9jTosF5zhe9hvGpKHZ4EEs0roA+oMp X-Received: from pfnh19.prod.google.com ([2002:a62:b413:0:b0:730:9951:c9ea]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a00:a02:b0:736:9f20:a175 with SMTP id d2e1a72fcca58-739059417ddmr25269223b3a.2.1742947729004; Tue, 25 Mar 2025 17:08:49 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:23 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-26-irogers@google.com> Subject: [PATCH v3 25/35] perf vendor events: Update rocketlake events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/rocketlake/cache.json | 60 +++ .../arch/x86/rocketlake/memory.json | 160 ++++++++ .../pmu-events/arch/x86/rocketlake/other.json | 220 ---------- .../arch/x86/rocketlake/rkl-metrics.json | 385 +++++++++--------- 4 files changed, 412 insertions(+), 413 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/rocketlake/cache.json b/tools/p= erf/pmu-events/arch/x86/rocketlake/cache.json index 791fa526d192..0f543325ec1a 100644 --- a/tools/perf/pmu-events/arch/x86/rocketlake/cache.json +++ b/tools/perf/pmu-events/arch/x86/rocketlake/cache.json @@ -445,6 +445,16 @@ "SampleAfterValue": "50021", "UMask": "0x20" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit a cacheline in the L3 where a snoop was s= ent or not.", "Counter": "0,1,2,3", @@ -505,6 +515,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit a cacheline= in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", @@ -565,6 +585,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit a = cacheline in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", @@ -625,6 +655,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that hit a cacheline in the L3 where a= snoop was sent or not.", "Counter": "0,1,2,3", @@ -655,6 +695,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that hit a cacheline in the L3 where a snoop was sent or n= ot.", "Counter": "0,1,2,3", @@ -715,6 +765,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that hit a cacheline in the L3 where a snoop was sent or not.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/rocketlake/memory.json b/tools/= perf/pmu-events/arch/x86/rocketlake/memory.json index abaf3f4f9d63..1455aaac37b1 100644 --- a/tools/perf/pmu-events/arch/x86/rocketlake/memory.json +++ b/tools/perf/pmu-events/arch/x86/rocketlake/memory.json @@ -176,6 +176,16 @@ "SampleAfterValue": "50021", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -186,6 +196,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that was not supplie= d by the L3 cache.", "Counter": "0,1,2,3", @@ -196,6 +226,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that was no= t supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -206,6 +256,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that was not supplied by the L3 cache.= ", "Counter": "0,1,2,3", @@ -216,6 +286,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -226,6 +316,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -236,6 +346,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.OTHER.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184008000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that was not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -246,6 +376,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.OTHER.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184008000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.STREAMING_WR.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000800", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that was not supplied= by the L3 cache.", "Counter": "0,1,2,3", @@ -256,6 +406,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", + "Counter": "0,1,2,3", + "EventCode": "0xB7, 0xBB", + "EventName": "OCR.STREAMING_WR.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000800", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/rocketlake/other.json b/tools/p= erf/pmu-events/arch/x86/rocketlake/other.json index a96b2a989d3f..141cd30a30af 100644 --- a/tools/perf/pmu-events/arch/x86/rocketlake/other.json +++ b/tools/perf/pmu-events/arch/x86/rocketlake/other.json @@ -26,186 +26,6 @@ "SampleAfterValue": "200003", "UMask": "0x20" }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that DRAM supplied t= he request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that DRAM s= upplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L1 data cache prefetch requests and so= ftware prefetches (except PREFETCHW) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L1D_AND_SWPF.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch data reads (which br= ing data to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetch RFOs (which bring da= ta to L2) that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that have any type of response.", "Counter": "0,1,2,3", @@ -216,26 +36,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.OTHER.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184008000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts miscellaneous requests, such as I/O an= d un-cacheable accesses that DRAM supplied the request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.OTHER.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184008000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -245,25 +45,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.STREAMING_WR.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000800", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts streaming stores that DRAM supplied th= e request.", - "Counter": "0,1,2,3", - "EventCode": "0xB7, 0xBB", - "EventName": "OCR.STREAMING_WR.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000800", - "SampleAfterValue": "100003", - "UMask": "0x1" } ] diff --git a/tools/perf/pmu-events/arch/x86/rocketlake/rkl-metrics.json b/t= ools/perf/pmu-events/arch/x86/rocketlake/rkl-metrics.json index cfda8956353e..71737a1a1997 100644 --- a/tools/perf/pmu-events/arch/x86/rocketlake/rkl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/rocketlake/rkl-metrics.json @@ -89,12 +89,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6) / (4 * tma_info_core_core_= clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -106,7 +106,7 @@ "MetricExpr": "34 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, @@ -129,12 +129,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -149,7 +149,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -157,7 +157,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store)) + tma_memory_bound * (tma_stor= e_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tm= a_store_bound)) * (tma_store_latency / (tma_store_latency + tma_false_shari= ng + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores)) + tma_memory_bound * (tma_stor= e_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tm= a_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_false_sharing = + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -165,22 +165,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_br= anches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma= _ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms /= (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_br= anches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tm= a_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms /= (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_mispredicts_resteers += tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_it= lb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_clears_resteers + tma_= mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_= dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -188,7 +188,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -196,15 +196,15 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears" + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -216,17 +216,17 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions.", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES= / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_branch_instructions", @@ -248,8 +248,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -257,8 +257,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -266,24 +266,24 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -291,7 +291,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -299,33 +299,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((32.5 * tma_info_system_core_frequency - 3.5 * tma= _info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM + (27 * tm= a_info_system_core_frequency - 3.5 * tma_info_system_core_frequency) * MEM_= LOAD_L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RE= TIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(29 * tma_info_system_core_frequency * MEM_LOAD_L3_= HIT_RETIRED.XSNP_HITM + 23.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L= 1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -335,25 +335,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(27 * tma_info_system_core_frequency - 3.5 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT * (1 + MEM_LOA= D_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "23.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _HIT_RETIRED.XSNP_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_= MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -362,7 +362,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -372,7 +372,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -382,7 +382,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -390,26 +390,26 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -417,8 +417,8 @@ "MetricExpr": "32.5 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, { @@ -437,7 +437,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -447,7 +447,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -465,7 +465,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -474,15 +474,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -490,7 +490,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -499,7 +499,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -508,7 +508,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -517,7 +517,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -526,7 +526,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -538,17 +538,17 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D0x1@) / IDQ.MITE_UOPS", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D1@) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -556,8 +556,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -569,28 +569,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -619,7 +619,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -628,7 +628,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -637,10 +637,11 @@ { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -701,11 +702,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -718,20 +719,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -773,7 +774,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -791,7 +792,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -799,7 +800,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -807,7 +808,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -815,7 +816,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -823,7 +824,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -831,7 +832,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -886,7 +887,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 11", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1011,7 +1012,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1073,8 +1074,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1101,12 +1102,12 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1147,14 +1148,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1195,7 +1195,7 @@ "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_= clks", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1203,7 +1203,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1211,7 +1211,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1239,7 +1239,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1248,15 +1248,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1266,13 +1265,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1288,14 +1287,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 7.5" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1303,8 +1302,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1312,7 +1311,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1321,7 +1320,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1331,7 +1330,7 @@ "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + L1D_PEND_MISS.FB_FULL_PERIODS)= * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1340,7 +1339,7 @@ "MetricExpr": "3.5 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1350,17 +1349,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(12.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETI= RED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "9 * tma_info_system_core_frequency * (MEM_LOAD_RETI= RED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) = / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1368,18 +1367,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1396,7 +1395,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1404,31 +1403,31 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1437,7 +1436,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1447,7 +1446,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%" }, { @@ -1457,15 +1456,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1474,7 +1473,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1485,11 +1484,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", @@ -1511,7 +1510,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1526,24 +1525,24 @@ }, { "BriefDescription": "This metric represents fraction of cycles whe= re (only) 4 uops were delivered by the MITE pipeline", - "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D0x4@ - cpu@IDQ.MITE_= UOPS\\,cmask\\=3D0x5@) / tma_info_thread_clks", + "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D4@ - cpu@IDQ.MITE_UO= PS\\,cmask\\=3D5@) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_gr= oup", "MetricName": "tma_mite_4wide", - "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_f= etch_bandwidth > 0.2", + "MetricThreshold": "tma_mite_4wide > 0.05 & (tma_mite > 0.1 & tma_= fetch_bandwidth > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D0x1@ / tma_info_core_co= re_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D1@ / tma_info_core_core= _clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1554,7 +1553,7 @@ "MetricExpr": "3 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1563,7 +1562,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -1578,19 +1577,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1634,8 +1633,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1643,8 +1642,8 @@ "MetricExpr": "cpu@EXE_ACTIVITY.3_PORTS_UTIL\\,umask\\=3D0x80@ / t= ma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1652,7 +1651,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1661,7 +1660,7 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= port_0, tma_port_1, tma_port_5, tma_port_6", "ScaleUnit": "100%" }, @@ -1670,14 +1669,14 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -1690,7 +1689,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1699,7 +1698,7 @@ "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clk= s", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUS= E_INST", "ScaleUnit": "100%" }, @@ -1709,7 +1708,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1718,8 +1717,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1727,7 +1726,7 @@ "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1736,8 +1735,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1746,8 +1745,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1755,8 +1754,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 10 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1773,7 +1772,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -1781,31 +1780,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1813,7 +1812,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -1822,7 +1821,7 @@ "MetricExpr": "10 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1831,8 +1830,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 3C3BE74040 for ; Wed, 26 Mar 2025 00:09:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947742; cv=none; b=jknFD3H19oe8LcMaK7mxW6Ouv/AusaJjf2wFF08oqeCwca1Fg4syGkYL6jJ10+T4iXibk1iSLk9IbCb6366FV9WhyMx+Wfd+ayaNOgXCZWUKdMlRHfNiFkZelvOwYhcGg+7HFQq0OdBWhAzj4UjbT4YPwNe/9lnOajLoDPF9qQY= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947742; c=relaxed/simple; bh=Lp8ulElygeIKQ2bznVf0xqu32MtL52Ef9WdWB4yntak=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=E98iKoM4f+cp0+orim2eCJYhw9KVtehQHpuGNiMHIOYPe+buZ0Ay0X9K/+/2yuGsuiKt+xaPrZcHiTOA9UuLOTQenEq4/TD1i83Rxgz6dcDbuP/bO5emjjfZ9XmKmOIRnHcyOPCqF1E4iDEQZqbSMrLHs5RfmjeWrVLykTT24gU= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=P0A8guVM; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="P0A8guVM" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff605a7a43so16288427a91.3 for ; Tue, 25 Mar 2025 17:09:00 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947739; x=1743552539; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=Rht91ZAFHO6cO1rQpcP3HoB9XpYfuaUL/MILI96bbSc=; b=P0A8guVMWbVDaLRVsIm1eed2mopVezm+KVYdEO9zWmC6DySu+XUjS5oEiRCShz1GZW lYq/BCxBvmYknmJuD6jw3N2W+7MzgzcTkcMow+GyR+cDZxERc+8NYQRu50NSQRchICwu ixfdCHpCWl8/gJOyplOGCU9TShJABSTsJR5WKvw3zHVo/IVQ006OI5vQ/WlaXtiZA2Sj nht95XEcJpSDo3Dp45f4BieTE37sRpZvDuullEbBWUuYGnE0NMsbll+kQPlyDtJLaz+n llUO6keFCQdcPye/51g2OYL1G0NjZ5evNqw93VPt9GboeuyPhMCreVcnC1psywzMBCZ0 OR/A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947739; x=1743552539; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=Rht91ZAFHO6cO1rQpcP3HoB9XpYfuaUL/MILI96bbSc=; b=qdGoqNzizvDia5v9Ylixc28uHhIbOY4e8QOq5VcAwf+fRy5hgg/DRwYaivXGzH1ZSF fNp0AGH5smgxrJCm1YixA1VaTArpzCbKjlSoZNi7NjvhCBJUD3YEoWxEoaYTXE749hfF PfFKjitbPOlpgHIGfyYdkTtZ6WE5UZz2sWYWes8EAOrTv5f9ZOzUlfsl4MjvSRDO+9HD LgzBc5SYfaYVt+NkiZloEtQ+w3KhIn72MmCpDZoG201POY0VVpiuUHKVA76UMUuJuMce lk3fCgmm6tTVGcWgmf4Oz3p5/ccs1CcI+Md2Bz55LHWEUwoBpRrdQ3A79pS6N1LZsEdi kOug== X-Forwarded-Encrypted: i=1; AJvYcCURwyj32qC2P9Mca6nIFymMB5kN97it3icMLoQQb21oox70zhUYOfjyqUgLLH+0bnroIlQ0Q8IX2tMLzF0=@vger.kernel.org X-Gm-Message-State: AOJu0YxyH1b3hJFItMDqOgC99lNACaAOyMcDodNmSy+PUbK7MaNd+nky WpovLVzbo1Lm+62qAmXwLLeacqOBOJINT42lhPmj9jZ8eC5olkzIwOjGGPPlHOMLJCP3+nobcQu NOzM6iA== X-Google-Smtp-Source: AGHT+IFxppHQqu9GHdVDF4fR7msdtYnmfWt2akuj4uxRezXo8vhDdLC0CyRMntd/QF4PBRJxyeZX3W8UB2hJ X-Received: from pjbpd9.prod.google.com ([2002:a17:90b:1dc9:b0:2fc:c98:ea47]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:2f44:b0:2ff:58a4:9db5 with SMTP id 98e67ed59e1d1-303100219ecmr28388064a91.30.1742947739578; Tue, 25 Mar 2025 17:08:59 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:24 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-27-irogers@google.com> Subject: [PATCH v3 26/35] perf vendor events: Update sandybridge metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update TMA metrics from 4.8 to 5.02. Move INSTS_WRITTEN_TO_IQ.INSTS to the frontend topic. Signed-off-by: Ian Rogers --- .../arch/x86/sandybridge/frontend.json | 8 +++++ .../arch/x86/sandybridge/metricgroups.json | 5 +++ .../arch/x86/sandybridge/other.json | 8 ----- .../arch/x86/sandybridge/snb-metrics.json | 36 ++++++++++++++----- 4 files changed, 41 insertions(+), 16 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/sandybridge/frontend.json b/too= ls/perf/pmu-events/arch/x86/sandybridge/frontend.json index e95d1005e22f..5c9ab7680762 100644 --- a/tools/perf/pmu-events/arch/x86/sandybridge/frontend.json +++ b/tools/perf/pmu-events/arch/x86/sandybridge/frontend.json @@ -278,5 +278,13 @@ "EventName": "IDQ_UOPS_NOT_DELIVERED.CYCLES_LE_3_UOP_DELIV.CORE", "SampleAfterValue": "2000003", "UMask": "0x1" + }, + { + "BriefDescription": "Valid instructions written to IQ per cycle.", + "Counter": "0,1,2,3", + "EventCode": "0x17", + "EventName": "INSTS_WRITTEN_TO_IQ.INSTS", + "SampleAfterValue": "2000003", + "UMask": "0x1" } ] diff --git a/tools/perf/pmu-events/arch/x86/sandybridge/metricgroups.json b= /tools/perf/pmu-events/arch/x86/sandybridge/metricgroups.json index 7dc7eb0d3dd3..eb8fbd14138a 100644 --- a/tools/perf/pmu-events/arch/x86/sandybridge/metricgroups.json +++ b/tools/perf/pmu-events/arch/x86/sandybridge/metricgroups.json @@ -9,6 +9,7 @@ "BvCB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvFB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvIO": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", + "BvMB": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvML": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMP": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", "BvMS": "Grouping from Top-down Microarchitecture Analysis Metrics spr= eadsheet", @@ -33,6 +34,7 @@ "InsType": "Grouping from Top-down Microarchitecture Analysis Metrics = spreadsheet", "L2Evicts": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "LSD": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", + "LockCont": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "MachineClears": "Grouping from Top-down Microarchitecture Analysis Me= trics spreadsheet", "Machine_Clears": "Grouping from Top-down Microarchitecture Analysis M= etrics spreadsheet", "Mem": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -48,6 +50,7 @@ "Pipeline": "Grouping from Top-down Microarchitecture Analysis Metrics= spreadsheet", "PortsUtil": "Grouping from Top-down Microarchitecture Analysis Metric= s spreadsheet", "Power": "Grouping from Top-down Microarchitecture Analysis Metrics sp= readsheet", + "Prefetches": "Grouping from Top-down Microarchitecture Analysis Metri= cs spreadsheet", "Ret": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", "Retire": "Grouping from Top-down Microarchitecture Analysis Metrics s= preadsheet", "SMT": "Grouping from Top-down Microarchitecture Analysis Metrics spre= adsheet", @@ -75,6 +78,7 @@ "tma_bad_speculation_group": "Metrics contributing to tma_bad_speculat= ion category", "tma_branch_resteers_group": "Metrics contributing to tma_branch_reste= ers category", "tma_core_bound_group": "Metrics contributing to tma_core_bound catego= ry", + "tma_divider_group": "Metrics contributing to tma_divider category", "tma_dram_bound_group": "Metrics contributing to tma_dram_bound catego= ry", "tma_dtlb_load_group": "Metrics contributing to tma_dtlb_load category= ", "tma_dtlb_store_group": "Metrics contributing to tma_dtlb_store catego= ry", @@ -99,6 +103,7 @@ "tma_issueSmSt": "Metrics related by the issue $issueSmSt", "tma_issueSyncxn": "Metrics related by the issue $issueSyncxn", "tma_issueTLB": "Metrics related by the issue $issueTLB", + "tma_itlb_misses_group": "Metrics contributing to tma_itlb_misses cate= gory", "tma_l1_bound_group": "Metrics contributing to tma_l1_bound category", "tma_light_operations_group": "Metrics contributing to tma_light_opera= tions category", "tma_machine_clears_group": "Metrics contributing to tma_machine_clear= s category", diff --git a/tools/perf/pmu-events/arch/x86/sandybridge/other.json b/tools/= perf/pmu-events/arch/x86/sandybridge/other.json index 42692fa24b6c..970839a9c786 100644 --- a/tools/perf/pmu-events/arch/x86/sandybridge/other.json +++ b/tools/perf/pmu-events/arch/x86/sandybridge/other.json @@ -33,14 +33,6 @@ "SampleAfterValue": "2000003", "UMask": "0x2" }, - { - "BriefDescription": "Valid instructions written to IQ per cycle.", - "Counter": "0,1,2,3", - "EventCode": "0x17", - "EventName": "INSTS_WRITTEN_TO_IQ.INSTS", - "SampleAfterValue": "2000003", - "UMask": "0x1" - }, { "BriefDescription": "Cycles when L1 and L2 are locked due to UC or= split lock.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/sandybridge/snb-metrics.json b/= tools/perf/pmu-events/arch/x86/sandybridge/snb-metrics.json index ff2e515c744a..823d8b7c4224 100644 --- a/tools/perf/pmu-events/arch/x86/sandybridge/snb-metrics.json +++ b/tools/perf/pmu-events/arch/x86/sandybridge/snb-metrics.json @@ -127,7 +127,7 @@ "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", - "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_UOPS", + "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, { @@ -211,7 +211,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_256b, tma_fp_vector_512b, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_6, tma_ports= _utilized_2", "ScaleUnit": "100%" }, { @@ -220,7 +220,7 @@ "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_512b, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_6, tma_ports= _utilized_2", "ScaleUnit": "100%" }, { @@ -240,7 +240,7 @@ "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences. ([ICL+] Note this may overcou= nt due to approximation using indirect events; [ADL+] .)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -275,6 +275,12 @@ "MetricThreshold": "tma_info_frontend_dsb_coverage < 0.7 & tma_inf= o_thread_ipc / 4 > 0.35", "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_lcp" }, + { + "BriefDescription": "Taken Branches retired Per Cycle", + "MetricExpr": "BR_INST_RETIRED.NEAR_TAKEN / tma_info_thread_clks", + "MetricGroup": "Branches;FetchBW", + "MetricName": "tma_info_frontend_tbpc" + }, { "BriefDescription": "Total number of retired Instructions", "MetricExpr": "INST_RETIRED.ANY", @@ -290,7 +296,7 @@ }, { "BriefDescription": "Measured Average Core Frequency for unhalted = processors [GHz]", - "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / dur= ation_time", + "MetricExpr": "tma_info_system_turbo_utilization * TSC / 1e9 / tma= _info_system_time", "MetricGroup": "Power;Summary", "MetricName": "tma_info_system_core_frequency" }, @@ -308,14 +314,14 @@ }, { "BriefDescription": "Average external Memory Bandwidth Use for rea= ds and writes [GB / sec]", - "MetricExpr": "64 * (UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_RE= QUESTS.ALL) / 1e6 / duration_time / 1e3", + "MetricExpr": "64 * (UNC_ARB_TRK_REQUESTS.ALL + UNC_ARB_COH_TRK_RE= QUESTS.ALL) / 1e6 / tma_info_system_time / 1e3", "MetricGroup": "HPC;MemOffcore;MemoryBW;SoC;tma_issueBW", "MetricName": "tma_info_system_dram_bw_use", "PublicDescription": "Average external Memory Bandwidth Use for re= ads and writes [GB / sec]. Related metrics: tma_mem_bandwidth" }, { "BriefDescription": "Giga Floating Point Operations Per Second", - "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / duration_time", + "MetricExpr": "(FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE + FP_COMP_OPS_EX= E.SSE_SCALAR_DOUBLE + 2 * FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE + 4 * (FP_COMP_= OPS_EXE.SSE_PACKED_SINGLE + SIMD_FP_256.PACKED_DOUBLE) + 8 * SIMD_FP_256.PA= CKED_SINGLE) / 1e9 / tma_info_system_time", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_system_gflops", "PublicDescription": "Giga Floating Point Operations Per Second. A= ggregate across all supported options of: FP precisions, scalar and vector = instructions, vector-width" @@ -340,6 +346,13 @@ "MetricName": "tma_info_system_kernel_utilization", "MetricThreshold": "tma_info_system_kernel_utilization > 0.05" }, + { + "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P / CPU_CLK_UNHALTED.THREAD= ", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" + }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", "MetricExpr": "(1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / (CPU_CLK_= UNHALTED.REF_XCLK_ANY / 2) if #SMT_on else 0)", @@ -352,6 +365,13 @@ "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, + { + "BriefDescription": "Run duration time in seconds", + "MetricExpr": "duration_time", + "MetricGroup": "Summary", + "MetricName": "tma_info_system_time", + "MetricThreshold": "tma_info_system_time < 1" + }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", "MetricExpr": "tma_info_thread_clks / CPU_CLK_UNHALTED.REF_TSC", @@ -448,7 +468,7 @@ { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D6@) / tma_info_thread_clks", - "MetricGroup": "BvMS;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", + "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_info_system_dram_bw_use", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 9321F17A2FC for ; Wed, 26 Mar 2025 00:09:09 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947760; cv=none; b=ittPnZKVUwz6kFgEvmHUsL3ATXRIA2rkD9H/M4HMLFqpS4n/4KKiNHu7nXIC+8+8Aeaz9sTc8u6XLxQnMdweoFrYmrZE9lOpiknaVe8xO8wPh0c8GpqdBVju5usCO/rZFInjW/ktPw1o8yrX27mIVKSrFPCWAQXV7+e+upy0aCg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947760; c=relaxed/simple; bh=FeR6hBQQa7dkRUZfqcb9WP1KtcyoB4PtVMfoM9MobAM=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=BquNqJt72ETKw7VMLNGcKBieEwLO2jZaoJxSyCltIHrXYuf7dIiC4jM7c+8JTwq0OQ4w5vURKp9I8qlnmRCiXcuvNeKf3EYWQ0VAF+qy3kLI4C9CIcTtofyd0+IF/5qgQ7yAnA3FpN4mY/g1SzH5Ogaj6i8hbLzifS0A0B2nCYQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=ohXaTtFB; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="ohXaTtFB" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff68033070so9922315a91.2 for ; Tue, 25 Mar 2025 17:09:09 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947749; x=1743552549; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=zOBHuwy/zhiHKrGsubhIMyKwsk98oTcSleIPQIVkOXQ=; b=ohXaTtFBOXlkFK/y/hy3KlckRJDmttiID9cyhOmpBGTksy42eCWct0rWlBI9gSGvkW LukVU2LwsK+Ah2p5GzCxrCRWJpYq7+iIFVL8mRulM6WpzZdBI7EqYkCqEq6AyobkAkyv E+DluzzkynB4u6E6v7afXfuO/Pa3x1h0e+n565AINpnDjVttCs679NLTfDo7ZGUS+8Xn gVbDY5PmDJ6bTsw3aSrWo2sOrWiQWa7+gmeWNs4QP3jF2vZq+JEfFzEMKWCni+3ocmqT H5ackIV2+lek858B14MNrExba65vAJQcl0w5aJNHESS/UwSbzG8r6eOJSb/DSMe5QmPC kyMA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947749; x=1743552549; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=zOBHuwy/zhiHKrGsubhIMyKwsk98oTcSleIPQIVkOXQ=; b=wREq4pFdgGBMGdHBSy8o39styGhkqi1Iv6jpPcmqgMPrTNSr2HtIpV3P9OXsiS1KkG hBBuqHtlftmfyqvjl0XvRR3UznipdTG4bVC4ngK0veYdzSHzo2MnsQmuK0zZFX4mLcdH KA7kk3UfKp9IEworj3rfvoevfnqa0qUc4K9Ot4x18qlvy0YUG1sB6G0kuEJoAgDcdFSc a6ZuFZHJuMQeAi+vvx5WOvpKegXt6FhufsSUNPBeUGDhTIivbZk7OgHudtXPyKNjSUVF LulfpS8tkxpj5qpUUrrUAxFz/DIEZ0ZcgSAOKDiVMjPJaokuI3jwWdTCaCmJ8bBt/TAi +lTQ== X-Forwarded-Encrypted: i=1; AJvYcCUjktKv9UZ9N9y/8cK1W7BjgTG73sL7w8C0nW2hU2FMx/Ev1hff/SKlSYCoR+qcErKKXAvrO5+QyPynNCE=@vger.kernel.org X-Gm-Message-State: AOJu0YyJJ74Zj7GQBnJpCQLIQMaqV+DPM8MzjXR9PSvF+CUHZHfXMKGK CB6o5QwIckdiW2FjuiHoRpG1BJg0wIUAasKKSOW8a1vnsY1+sMqD6G7/t3+noR309coQ2pZllmq kHTUBxQ== X-Google-Smtp-Source: AGHT+IGV8NQpkx/8sei8EBnlUmu4FmcasfITxzSj0cvpDZeT5i/MgeQB/1KP06KJpW0ouc1rgS07UShTNnwU X-Received: from pgbbq17.prod.google.com ([2002:a05:6a02:451:b0:af5:5fab:ca93]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:9188:b0:1f5:64fd:68eb with SMTP id adf61e73a8af0-1fe42f07ba9mr32989138637.7.1742947748879; Tue, 25 Mar 2025 17:09:08 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:25 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-28-irogers@google.com> Subject: [PATCH v3 27/35] perf vendor events: Update sapphirerapids events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../arch/x86/sapphirerapids/cache.json | 150 ++++++ .../arch/x86/sapphirerapids/memory.json | 170 +++++++ .../arch/x86/sapphirerapids/other.json | 378 -------------- .../arch/x86/sapphirerapids/pipeline.json | 58 +++ .../arch/x86/sapphirerapids/spr-metrics.json | 465 +++++++++--------- 5 files changed, 610 insertions(+), 611 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/sapphirerapids/cache.json b/too= ls/perf/pmu-events/arch/x86/sapphirerapids/cache.json index e35dbb7c2ccd..4363e53169f7 100644 --- a/tools/perf/pmu-events/arch/x86/sapphirerapids/cache.json +++ b/tools/perf/pmu-events/arch/x86/sapphirerapids/cache.json @@ -588,6 +588,16 @@ "SampleAfterValue": "1000003", "UMask": "0x3" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that hit in the L3 or were snooped from another co= re's caches on the same socket.", "Counter": "0,1,2,3", @@ -628,6 +638,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit in the L3 o= r were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -668,6 +688,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to this socket, whether or not in Sub NUMA Cluster(SNC) Mode= . In SNC Mode counts PMM accesses that are controlled by the close or dist= ant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_SOCKET_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700C00001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703C00001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y a cache on a remote socket where a snoop hit a modified line in another c= ore's caches which forwarded the data.", "Counter": "0,1,2,3", @@ -688,6 +728,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that hit a modified = line in a distant L3 Cache or were snooped from a distant core's L1/L2 cach= es on this socket when the system is in SNC (sub-NUMA cluster) mode.", "Counter": "0,1,2,3", @@ -708,6 +758,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC0002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that hit in= the L3 or were snooped from another core's caches on the same socket.", "Counter": "0,1,2,3", @@ -748,6 +808,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts data load hardware prefetch requests t= o the L1 data cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L1D.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetches (which bring data = to L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L2.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10070", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L3.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x12380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t hit in the L3 or were snooped from another core's caches on the same sock= et.", "Counter": "0,1,2,3", @@ -758,6 +848,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.HWPF_L3.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x90002380", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10808", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F3FFC4477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit in the L3 or were snooped from another core's caches on the sa= me socket.", "Counter": "0,1,2,3", @@ -798,6 +918,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, whether or not in Su= b NUMA Cluster(SNC) Mode. In SNC Mode counts PMM accesses that are control= led by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x700C04477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3F33004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by a cache on a remote socket where a snoop was sent= and data was returned (Modified or Not Modified).", "Counter": "0,1,2,3", @@ -828,6 +968,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_PMM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x703004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that hit a modified line in a distant L3 Cache or were snooped from a d= istant core's L1/L2 caches on this socket when the system is in SNC (sub-NU= MA cluster) mode.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/sapphirerapids/memory.json b/to= ols/perf/pmu-events/arch/x86/sapphirerapids/memory.json index 41d4120d4dae..981e573330cd 100644 --- a/tools/perf/pmu-events/arch/x86/sapphirerapids/memory.json +++ b/tools/perf/pmu-events/arch/x86/sapphirerapids/memory.json @@ -173,6 +173,16 @@ "SampleAfterValue": "1000003", "UMask": "0x2" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the local socket's L1, L= 2, or L3 caches.", "Counter": "0,1,2,3", @@ -183,6 +193,36 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were not suppli= ed by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -193,6 +233,46 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were n= ot supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -203,6 +283,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.DEMAND_RFO.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts hardware prefetches to the L3 only tha= t missed the local socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -223,6 +323,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x73C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches.", "Counter": "0,1,2,3", @@ -253,6 +363,56 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x104004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x70C004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x733004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.READS_TO_CORE.SNC_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x708004477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that missed the local= socket's L1, L2, and L3 caches.", "Counter": "0,1,2,3", @@ -273,6 +433,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", + "Counter": "0,1,2,3", + "EventCode": "0x2A,0x2B", + "EventName": "OCR.WRITE_ESTIMATE.MEMORY", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0xFBFF80822", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data read requests that miss th= e L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/sapphirerapids/other.json b/too= ls/perf/pmu-events/arch/x86/sapphirerapids/other.json index 05d8f14956ee..df4019ff7883 100644 --- a/tools/perf/pmu-events/arch/x86/sapphirerapids/other.json +++ b/tools/perf/pmu-events/arch/x86/sapphirerapids/other.json @@ -7,324 +7,6 @@ "SampleAfterValue": "1000003", "UMask": "0x8" }, - { - "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xb7", - "EventName": "EXE.AMX_BUSY", - "SampleAfterValue": "2000003", - "UMask": "0x2" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM attached to this socket= , unless in Sub NUMA Cluster(SNC) Mode. In SNC Mode counts only those DRAM= accesses that are controlled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM on a distant memory con= troller of this socket when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_CODE_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mode. In S= NC Mode counts only those DRAM accesses that are controlled by the close SN= C Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to this socket, whether or not in Sub NUMA Cluster(SNC) Mode= . In SNC Mode counts PMM accesses that are controlled by the close or dist= ant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_SOCKET_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700C00001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703C00001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM on a distant memory controller of this socket when the system is in = SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_DATA_RD.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that have a= ny type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC0002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM attached to this socket, unless in Sub NUMA Cluster(SNC) Mo= de. In SNC Mode counts only those DRAM accesses that are controlled by the= close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) reque= sts and software prefetches for exclusive ownership (PREFETCHW) that were s= upplied by DRAM on a distant memory controller of this socket when the syst= em is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.DEMAND_RFO.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts data load hardware prefetch requests t= o the L1 data cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L1D.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches (which bring data = to L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L2.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10070", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L3.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x12380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts hardware prefetches to the L3 only tha= t were not supplied by the local socket's L1, L2, or L3 caches and the cach= eline was homed in a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.HWPF_L3.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x90002380", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts writebacks of modified cachelines and = streaming stores that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.MODIFIED_WRITE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10808", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F3FFC4477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x73C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, unless in Sub NUMA = Cluster(SNC) Mode. In SNC Mode counts only those DRAM accesses that are co= ntrolled by the close SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x104004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to this socket, whether or not in S= ub NUMA Cluster(SNC) Mode. In SNC Mode counts DRAM accesses that are contr= olled by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x70C004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to this socket, whether or not in Su= b NUMA Cluster(SNC) Mode. In SNC Mode counts PMM accesses that are control= led by the close or distant SNC Cluster.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.LOCAL_SOCKET_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x700C04477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were not supplied by the local socket's L1, L2, or L3 caches and w= ere supplied by a remote socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3F33004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM or PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x733004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by PMM attached to another socket.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.REMOTE_PMM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x703004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all (cacheable) data read, code read a= nd RFO requests including demands and prefetches to the core caches (L1 or = L2) that were supplied by DRAM on a distant memory controller of this socke= t when the system is in SNC (sub-NUMA cluster) mode.", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.READS_TO_CORE.SNC_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x708004477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -335,66 +17,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts Demand RFOs, ItoM's, PREFECTHW's, Hard= ware RFO Prefetches to the L1/L2 and Streaming stores that likely resulted = in a store to Memory (DRAM or PMM)", - "Counter": "0,1,2,3", - "EventCode": "0x2A,0x2B", - "EventName": "OCR.WRITE_ESTIMATE.MEMORY", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0xFBFF80822", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY", - "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", - "SampleAfterValue": "1000003", - "UMask": "0x7" - }, - { - "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_COUNT", - "Invert": "1", - "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", - "SampleAfterValue": "100003", - "UMask": "0x7" - }, - { - "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xa5", - "EventName": "RS.EMPTY_RESOURCE", - "SampleAfterValue": "1000003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", - "Counter": "0,1,2,3,4,5,6,7", - "CounterMask": "1", - "Deprecated": "1", - "EdgeDetect": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.COUNT", - "Invert": "1", - "SampleAfterValue": "100003", - "UMask": "0x7" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", - "Counter": "0,1,2,3,4,5,6,7", - "Deprecated": "1", - "EventCode": "0xa5", - "EventName": "RS_EMPTY.CYCLES", - "SampleAfterValue": "1000003", - "UMask": "0x7" - }, { "BriefDescription": "Cycles the uncore cannot take further request= s", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/sapphirerapids/pipeline.json b/= tools/perf/pmu-events/arch/x86/sapphirerapids/pipeline.json index 50cacfbbc7cf..c16b63979c55 100644 --- a/tools/perf/pmu-events/arch/x86/sapphirerapids/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/sapphirerapids/pipeline.json @@ -367,6 +367,14 @@ "SampleAfterValue": "1000003", "UMask": "0x4" }, + { + "BriefDescription": "Counts the cycles where the AMX (Advance Matr= ix Extension) unit is busy performing an operation.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xb7", + "EventName": "EXE.AMX_BUSY", + "SampleAfterValue": "2000003", + "UMask": "0x2" + }, { "BriefDescription": "Cycles total of 1 uop is executed on all port= s and Reservation Station was not empty.", "Counter": "0,1,2,3,4,5,6,7", @@ -740,6 +748,56 @@ "SampleAfterValue": "100003", "UMask": "0x2" }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= for the thread.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY", + "PublicDescription": "Counts cycles during which the reservation s= tation (RS) is empty for this logical processor. This is usually caused whe= n the front-end pipeline runs into starvation periods (e.g. branch mispredi= ctions or i-cache misses)", + "SampleAfterValue": "1000003", + "UMask": "0x7" + }, + { + "BriefDescription": "Counts end of periods where the Reservation S= tation (RS) was empty.", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_COUNT", + "Invert": "1", + "PublicDescription": "Counts end of periods where the Reservation = Station (RS) was empty. Could be useful to closely sample on front-end late= ncy issues (see the FRONTEND_RETIRED event of designated precise events)", + "SampleAfterValue": "100003", + "UMask": "0x7" + }, + { + "BriefDescription": "Cycles when Reservation Station (RS) is empty= due to a resource in the back-end", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xa5", + "EventName": "RS.EMPTY_RESOURCE", + "SampleAfterValue": "1000003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY_COUNT", + "Counter": "0,1,2,3,4,5,6,7", + "CounterMask": "1", + "Deprecated": "1", + "EdgeDetect": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.COUNT", + "Invert": "1", + "SampleAfterValue": "100003", + "UMask": "0x7" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = RS.EMPTY", + "Counter": "0,1,2,3,4,5,6,7", + "Deprecated": "1", + "EventCode": "0xa5", + "EventName": "RS_EMPTY.CYCLES", + "SampleAfterValue": "1000003", + "UMask": "0x7" + }, { "BriefDescription": "TMA slots where no uops were being issued due= to lack of back-end resources.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/sapphirerapids/spr-metrics.json= b/tools/perf/pmu-events/arch/x86/sapphirerapids/spr-metrics.json index b59fae4a887d..fc87899a2168 100644 --- a/tools/perf/pmu-events/arch/x86/sapphirerapids/spr-metrics.json +++ b/tools/perf/pmu-events/arch/x86/sapphirerapids/spr-metrics.json @@ -360,7 +360,7 @@ "ScaleUnit": "1per_instr" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5_11 + UOPS_DISPATCHED.PORT_6) / (5 * tma_info_core_co= re_clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -372,7 +372,7 @@ "MetricExpr": "EXE.AMX_BUSY / tma_info_core_core_clks", "MetricGroup": "BvCB;Compute;HPC;Server;TopdownL3;tma_L3_group;tma= _core_bound_group", "MetricName": "tma_amx_busy", - "MetricThreshold": "tma_amx_busy > 0.5 & tma_core_bound > 0.1 & tm= a_backend_bound > 0.2", + "MetricThreshold": "tma_amx_busy > 0.5 & (tma_core_bound > 0.1 & t= ma_backend_bound > 0.2)", "ScaleUnit": "100%" }, { @@ -380,12 +380,12 @@ "MetricExpr": "78 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists", + "BriefDescription": "This metric estimates fraction of slots the C= PU retired uops as a result of handing SSE to AVX* or AVX* to SSE transitio= n Assists.", "MetricExpr": "63 * ASSISTS.SSE_AVX_MIX / tma_info_thread_slots", "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_avx_assists", @@ -395,7 +395,7 @@ { "BriefDescription": "This category represents fraction of slots wh= ere no uops are being delivered due to a lack of required resources for acc= epting new uops in the Backend", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-be\\-bound / (topdown\\-fe\\-bound + topd= own\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_inf= o_thread_slots", "MetricGroup": "BvOB;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", @@ -411,12 +411,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -431,7 +431,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_fb_full)= ))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_fb_full + tma_l1_l= atency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)= ))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -439,7 +439,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_= bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_d= ram_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_st= ore_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_load= s + tma_fb_full)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_= l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_fb_full)) + tma_memory_bound * (tma_s= tore_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_split_stores / (tma_store_latency + tma_false_sha= ring + tma_split_stores + tma_streaming_stores + tma_dtlb_store)) + tma_mem= ory_bound * (tma_store_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound = + tma_dram_bound + tma_store_bound)) * (tma_store_latency / (tma_store_late= ncy + tma_false_sharing + tma_split_stores + tma_streaming_stores + tma_dtl= b_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependen= cy + tma_lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_= bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma= _l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_dtlb_load + tma_fb= _full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tm= a_store_fwd_blk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tm= a_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_split_l= oads / (tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_= latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_s= tore_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_split_stores / (tma_dtlb_store + tma_false_sharin= g + tma_split_stores + tma_store_latency + tma_streaming_stores)) + tma_mem= ory_bound * (tma_store_bound / (tma_dram_bound + tma_l1_bound + tma_l2_boun= d + tma_l3_bound + tma_store_bound)) * (tma_store_latency / (tma_dtlb_store= + tma_false_sharing + tma_split_stores + tma_store_latency + tma_streaming= _stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -447,22 +447,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_amx_busy + tma_ports_utilization) + tma_c= ore_bound * tma_amx_busy / (tma_divider + tma_serializing_operation + tma_a= mx_busy + tma_ports_utilization) + tma_core_bound * (tma_ports_utilization = / (tma_divider + tma_serializing_operation + tma_amx_busy + tma_ports_utili= zation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_amx_busy= + tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_c= ore_bound * tma_amx_busy / (tma_amx_busy + tma_divider + tma_ports_utilizat= ion + tma_serializing_operation) + tma_core_bound * (tma_ports_utilization = / (tma_amx_busy + tma_divider + tma_ports_utilization + tma_serializing_ope= ration)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_utili= zed_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1@) * (tma_fe= tch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers= + tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredict= s) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_branches= )) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_sw= itches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms / (tma_= mite + tma_dsb + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - (1 - I= NST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.MS\\,cmask\\=3D1@) * (tma_fetc= h_latency * (tma_ms_switches + tma_branch_resteers * (tma_clears_resteers += tma_mispredicts_resteers * tma_other_mispredicts / tma_branch_mispredicts)= / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_branches))= / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_m= isses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms / (tma_ds= b + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D0x1@) * (tma_fetch_latency * (tma_ms_switches + tma_b= ranch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_othe= r_mispredicts / tma_branch_mispredicts) / (tma_mispredicts_resteers + tma_c= lears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_itlb_mis= ses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) += tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_ms)) + 10 * tma_m= icrocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_b= ranch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes = + tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_inf= o_thread_clks * tma_ports_utilized_0) / (tma_divider + tma_serializing_oper= ation + tma_amx_busy + tma_ports_utilization) + tma_microcode_sequencer / (= tma_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_m= icrocode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * ((1 - INST_RETIRED.REP_ITERATION / cpu@UOPS_R= ETIRED.MS\\,cmask\\=3D1@) * (tma_fetch_latency * (tma_ms_switches + tma_bra= nch_resteers * (tma_clears_resteers + tma_mispredicts_resteers * tma_other_= mispredicts / tma_branch_mispredicts) / (tma_clears_resteers + tma_mispredi= cts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_dsb_swit= ches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) + t= ma_fetch_bandwidth * tma_ms / (tma_dsb + tma_mite + tma_ms)) + 10 * tma_mic= rocode_sequencer * tma_other_mispredicts / tma_branch_mispredicts * tma_bra= nch_mispredicts + tma_machine_clears * tma_other_nukes / tma_other_nukes + = tma_core_bound * (tma_serializing_operation + RS.EMPTY_RESOURCE / tma_info_= thread_clks * tma_ports_utilized_0) / (tma_amx_busy + tma_divider + tma_por= ts_utilization + tma_serializing_operation) + tma_microcode_sequencer / (tm= a_few_uops_instructions + tma_microcode_sequencer) * (tma_assists / tma_mic= rocode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -470,7 +470,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound += tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_store_latency + tma_false_sharing + tma_split_stores + tma_= streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dram_bound= + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_dt= lb_store / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tma_sto= re_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -478,7 +478,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency + tma_streaming_stores - tma_store_latency)) + t= ma_machine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -486,7 +486,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -498,10 +498,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -510,7 +510,7 @@ { "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Branch Misprediction", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * sl= ots", + "MetricExpr": "topdown\\-br\\-mispredict / (topdown\\-fe\\-bound += topdown\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tm= a_info_thread_slots", "MetricGroup": "BadSpec;BrMispredicts;BvMP;Default;TmaL2;TopdownL2= ;tma_L2_group;tma_bad_speculation_group;tma_issueBM", "MetricName": "tma_branch_mispredicts", "MetricThreshold": "tma_branch_mispredicts > 0.1 & tma_bad_specula= tion > 0.15", @@ -523,24 +523,24 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.1 power-performance optimized state (Fas= ter wakeup time; Smaller power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C01 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c01_wait", - "MetricThreshold": "tma_c01_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c01_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings)", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due staying in C0.2 power-performance optimized state (Slo= wer wakeup time; Larger power savings).", "MetricExpr": "CPU_CLK_UNHALTED.C02 / tma_info_thread_clks", "MetricGroup": "C0Wait;TopdownL4;tma_L4_group;tma_serializing_oper= ation_group", "MetricName": "tma_c02_wait", - "MetricThreshold": "tma_c02_wait > 0.05 & tma_serializing_operatio= n > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_c02_wait > 0.05 & (tma_serializing_operati= on > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -548,7 +548,7 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources. Sample with: FRONTEND_RETIRE= D.MS_FLOWS", "ScaleUnit": "100%" }, @@ -557,24 +557,24 @@ "MetricExpr": "(1 - tma_branch_mispredicts / tma_bad_speculation) = * INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -582,7 +582,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -590,32 +590,32 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", - "MetricExpr": "((81 * tma_info_system_core_frequency - 4.4 * tma_i= nfo_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (OCR.DEMAN= D_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) + (79 * tma_info_system_core_fre= quency - 4.4 * tma_info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XS= NP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / t= ma_info_thread_clks", + "MetricExpr": "(76.6 * tma_info_system_core_frequency * (MEM_LOAD_= L3_HIT_RETIRED.XSNP_FWD * (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMA= ND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD= ))) + 74.6 * tma_info_system_core_frequency * MEM_LOAD_L3_HIT_RETIRED.XSNP_= MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_= info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -626,24 +626,24 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", - "MetricExpr": "(79 * tma_info_system_core_frequency - 4.4 * tma_in= fo_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD= _L3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR= .DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WIT= H_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / t= ma_info_thread_clks", + "MetricExpr": "74.6 * tma_info_system_core_frequency * (MEM_LOAD_L= 3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEM= AND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR= .DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT= / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -652,7 +652,7 @@ "MetricExpr": "ARITH.DIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -661,7 +661,7 @@ "MetricExpr": "MEMORY_ACTIVITY.STALLS_L3_MISS / tma_info_thread_cl= ks", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -671,7 +671,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -679,34 +679,34 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEM= ORY_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - MEMOR= Y_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates how often CPU w= as handling synchronizations due to False Sharing", - "MetricExpr": "(170 * tma_info_system_core_frequency * cpu@OCR.DEM= AND_RFO.L3_MISS\\,offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core= _frequency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", + "MetricExpr": "(170 * tma_info_system_core_frequency * OCR.DEMAND_= RFO.L3_MISS@offcore_rsp\\=3D0x103b800002@ + 81 * tma_info_system_core_frequ= ency * OCR.DEMAND_RFO.L3_HIT.SNOOP_HITM) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, @@ -727,7 +727,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -738,7 +738,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -756,7 +756,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -765,15 +765,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FPDIV_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -781,8 +781,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED2.SCALAR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_vector_2= 56b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -790,8 +790,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.VECTOR + FP_ARITH_INST_RETIR= ED2.VECTOR) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6= , tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -799,8 +799,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.128B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -808,8 +808,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.256B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized= _2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_int_vector_128b, = tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_po= rts_utilized_2", "ScaleUnit": "100%" }, { @@ -817,8 +817,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE + FP_ARITH_INST_RETIRED2.512B_PACKED_HALF= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_int_vector_128b, tma_int_vecto= r_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_= 2", "ScaleUnit": "100%" }, { @@ -829,27 +829,27 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * INST_RETIRED.MACRO_FUSED / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-heavy\\-ops / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+]). Sample with: UOPS_RET= IRED.HEAVY", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+]). Sample with: UOPS_RE= TIRED.HEAVY", "ScaleUnit": "100%" }, { @@ -857,8 +857,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -869,28 +869,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -918,7 +918,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -926,7 +926,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -934,10 +934,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -998,11 +999,11 @@ "MetricExpr": "(FP_ARITH_DISPATCHED.PORT_0 + FP_ARITH_DISPATCHED.P= ORT_1 + FP_ARITH_DISPATCHED.PORT_5) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -1015,20 +1016,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask= \\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_DATA.STALLS / cpu@ICACHE_DATA.STALLS\\,cmask= \\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -1065,13 +1066,13 @@ }, { "BriefDescription": "Average number of cycles the front-end was de= layed due to an Unknown Branch detection", - "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / cpu@INT_MISC.UNKNO= WN_BRANCH_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed", "MetricName": "tma_info_frontend_unknown_branch_cost", - "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node" + "PublicDescription": "Average number of cycles the front-end was d= elayed due to an Unknown Branch detection. See Unknown_Branches node." }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -1089,7 +1090,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -1097,7 +1098,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -1105,7 +1106,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -1113,7 +1114,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -1121,7 +1122,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Half-Pr= ecision instruction (lower number means higher occurrence rate)", @@ -1129,7 +1130,7 @@ "MetricGroup": "Flops;FpScalar;InsType;Server", "MetricName": "tma_info_inst_mix_iparith_scalar_hp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_hp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Half-P= recision instruction (lower number means higher occurrence rate). Values < = 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -1137,7 +1138,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1192,7 +1193,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 6 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 13", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1329,7 +1330,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1394,21 +1395,21 @@ "MetricExpr": "64 * OCR.READS_TO_CORE.DRAM / 1e9 / tma_info_system= _time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_dram_bw", - "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW" + "PublicDescription": "Average DRAM BW for Reads-to-Core (R2C) cove= ring for memory attached to local- and remote-socket. See R2C_Offcore_BW." }, { "BriefDescription": "Average L3-cache miss BW for Reads-to-Core (R= 2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.L3_MISS / 1e9 / tma_info_sys= tem_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_l3m_bw", - "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW" + "PublicDescription": "Average L3-cache miss BW for Reads-to-Core (= R2C). This covering going to DRAM or other memory off-chip memory tears. Se= e R2C_Offcore_BW." }, { "BriefDescription": "Average Off-core access BW for Reads-to-Core = (R2C)", "MetricExpr": "64 * OCR.READS_TO_CORE.ANY_RESPONSE / 1e9 / tma_inf= o_system_time", "MetricGroup": "HPC;Mem;MemoryBW;SoC", "MetricName": "tma_info_memory_soc_r2c_offcore_bw", - "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches" + "PublicDescription": "Average Off-core access BW for Reads-to-Core= (R2C). R2C account for demand or prefetch load/RFO/code access that fill d= ata into the Core caches." }, { "BriefDescription": "STLB (2nd level TLB) code speculative misses = per kilo instruction (misses of any page-size that complete the page walk)", @@ -1436,8 +1437,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1458,18 +1459,18 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, { "BriefDescription": "Estimated fraction of retirement-cycles deali= ng with repeat instructions", - "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D0x1@", + "MetricExpr": "INST_RETIRED.REP_ITERATION / cpu@UOPS_RETIRED.SLOTS= \\,cmask\\=3D1@", "MetricGroup": "MicroSeq;Pipeline;Ret", "MetricName": "tma_info_pipeline_strings_cycles", "MetricThreshold": "tma_info_pipeline_strings_cycles > 0.1" @@ -1532,14 +1533,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1550,7 +1550,7 @@ }, { "BriefDescription": "Average latency of data read request to exter= nal DRAM memory [in nanoseconds]", - "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / cha_0@event\\=3D0x0@", + "MetricExpr": "1e9 * (UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_= CHA_TOR_INSERTS.IA_MISS_DRD_DDR) / uncore_cha_0@event\\=3D0x1@", "MetricGroup": "MemOffcore;MemoryLat;Server;SoC", "MetricName": "tma_info_system_mem_dram_read_latency", "PublicDescription": "Average latency of data read request to exte= rnal DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data= -read prefetches" @@ -1560,11 +1560,11 @@ "MetricExpr": "UNC_CHA_RxC_IRQ1_REJECT.PA_MATCH / UNC_CHA_CLOCKTIC= KS", "MetricGroup": "LockCont;MemOffcore;Server;SoC", "MetricName": "tma_info_system_mem_irq_duplicate_address", - "MetricThreshold": "(tma_info_system_mem_irq_duplicate_address > 0= .1)" + "MetricThreshold": "tma_info_system_mem_irq_duplicate_address > 0.= 1" }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1598,7 +1598,7 @@ }, { "BriefDescription": "Socket actual clocks when any core is active = on that socket", - "MetricExpr": "cha_0@event\\=3D0x0@", + "MetricExpr": "uncore_cha_0@event\\=3D0x1@", "MetricGroup": "SoC", "MetricName": "tma_info_system_socket_clks" }, @@ -1628,7 +1628,7 @@ "MetricName": "tma_info_system_upi_data_transmit_bw" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1637,15 +1637,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1655,13 +1654,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1677,14 +1676,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 6 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 9" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1693,7 +1692,7 @@ "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_int_operations", "MetricThreshold": "tma_int_operations > 0.1 & tma_light_operation= s > 0.6", - "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain", + "PublicDescription": "This metric represents overall Integer (Int)= select operations fraction the CPU has executed (retired). Vector/Matrix I= nt operations and shuffles are counted. Note this metric's value may exceed= its parent due to use of \"Uops\" CountDomain.", "ScaleUnit": "100%" }, { @@ -1701,8 +1700,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_128 + INT_VEC_RETIRED.VNNI_128= ) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_128b", - "MetricThreshold": "tma_int_vector_128b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_128b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 128-bit vector Intege= r ADD/SUB/SAD or VNNI (Vector Neural Network Instructions) uops fraction th= e CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_ve= ctor_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_256b, tma= _port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1710,8 +1709,8 @@ "MetricExpr": "(INT_VEC_RETIRED.ADD_256 + INT_VEC_RETIRED.MUL_256 = + INT_VEC_RETIRED.VNNI_256) / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Compute;IntVector;Pipeline;TopdownL4;tma_L4_group;= tma_int_operations_group;tma_issue2P", "MetricName": "tma_int_vector_256b", - "MetricThreshold": "tma_int_vector_256b > 0.1 & tma_int_operations= > 0.1 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_int_vector_256b > 0.1 & (tma_int_operation= s > 0.1 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents 256-bit vector Intege= r ADD/SUB/SAD/MUL or VNNI (Vector Neural Network Instructions) uops fractio= n the CPU has retired. Related metrics: tma_fp_scalar, tma_fp_vector, tma_f= p_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b,= tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1719,8 +1718,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1728,7 +1727,7 @@ "MetricExpr": "max((EXE_ACTIVITY.BOUND_ON_LOADS - MEMORY_ACTIVITY.= STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1737,7 +1736,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - MEMORY_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1746,7 +1745,7 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L1D_MISS - MEMORY_ACTIVITY.= STALLS_L2_MISS) / tma_info_thread_clks", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1755,7 +1754,7 @@ "MetricExpr": "4.4 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1764,17 +1763,17 @@ "MetricExpr": "(MEMORY_ACTIVITY.STALLS_L2_MISS - MEMORY_ACTIVITY.S= TALLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(37 * tma_info_system_core_frequency - 4.4 * tma_in= fo_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETIRE= D.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "32.6 * tma_info_system_core_frequency * (MEM_LOAD_R= ETIRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2= )) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1782,19 +1781,19 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "DefaultMetricgroupName": "TopdownL2", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Default;Retire;TmaL2;TopdownL2;tma_L2_group;tma_re= tiring_group", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1811,7 +1810,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1819,39 +1818,39 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from local memory", - "MetricExpr": "(109 * tma_info_system_core_frequency - 37 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * (1 + MEM_= LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "72 * tma_info_system_core_frequency * MEM_LOAD_L3_M= ISS_RETIRED.LOCAL_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1= _MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1860,7 +1859,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1876,19 +1875,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling)", + "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to memory bandwidth Allocation= feature (RDT's memory bandwidth throttling).", "MetricExpr": "INT_MISC.MBA_STALLS / tma_info_thread_clks", "MetricGroup": "MemoryBW;Offcore;Server;TopdownL5;tma_L5_group;tma= _mem_bandwidth_group", "MetricName": "tma_mba_stalls", - "MetricThreshold": "tma_mba_stalls > 0.1 & tma_mem_bandwidth > 0.2= & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mba_stalls > 0.1 & (tma_mem_bandwidth > 0.= 2 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0= .2)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1897,32 +1896,32 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of slots the = Memory subsystem within the Backend was a bottleneck", "DefaultMetricgroupName": "TopdownL2", - "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-mem\\-bound / (topdown\\-fe\\-bound + top= down\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_in= fo_thread_slots", "MetricGroup": "Backend;Default;TmaL2;TopdownL2;tma_L2_group;tma_b= ackend_bound_group", "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2;Default", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions", + "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to LFENCE Instructions.", "MetricConstraint": "NO_GROUP_EVENTS_NMI", "MetricExpr": "13 * MISC2_RETIRED.LFENCE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_memory_fence", - "MetricThreshold": "tma_memory_fence > 0.05 & tma_serializing_oper= ation > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_memory_fence > 0.05 & (tma_serializing_ope= ration > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_UOP_RETIRED.ANY / (tma_r= etiring * tma_info_thread_slots)", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1943,7 +1942,7 @@ "MetricExpr": "tma_branch_mispredicts / tma_bad_speculation * INT_= MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1957,17 +1956,17 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "160 * ASSISTS.SSE_AVX_MIX / tma_info_thread_clks", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clk= s / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "max(IDQ.MS_CYCLES_ANY, cpu@UOPS_RETIRED.MS\\,cmask\= \=3D1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY)) / tma_info_core_core_clks = / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1975,10 +1974,10 @@ }, { "BriefDescription": "This metric estimates the fraction of cycles = when the CPU was stalled due to switches of uop delivery to the Microcode S= equencer (MS)", - "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D0x1\\,edge\\=3D= 0x1@ / (UOPS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", + "MetricExpr": "3 * cpu@UOPS_RETIRED.MS\\,cmask\\=3D1\\,edge@ / (UO= PS_RETIRED.SLOTS / UOPS_ISSUED.ANY) / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: FRONTEND_RETIRED.MS_FLOWS. Related metrics: tm= a_bottleneck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_mac= hine_clears, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_o= peration", "ScaleUnit": "100%" }, @@ -1988,7 +1987,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1996,7 +1995,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -2010,19 +2009,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -2031,7 +2030,7 @@ "MetricGroup": "TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_page_faults", "MetricThreshold": "tma_page_faults > 0.05", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Page Faults. A Page Fault m= ay apply on first application access to a memory page. Note operating syste= m handling of page faults accounts for the majority of its cost.", "ScaleUnit": "100%" }, { @@ -2040,7 +2039,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_6, tma_po= rts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_int_vector_128b, tma_int_vector_256b, tma_port_1, tma_port_5, tma_po= rt_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2049,7 +2048,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_int_vector_128b, tma_int_ve= ctor_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2058,7 +2057,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= ts_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_por= t_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -2066,8 +2065,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_3_PORTS_UTIL)) / tm= a_info_thread_clks if ARITH.DIV_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - EXE_= ACTIVITY.BOUND_ON_LOADS else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring * EX= E_ACTIVITY.2_3_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -2075,8 +2074,8 @@ "MetricExpr": "(EXE_ACTIVITY.EXE_BOUND_0_PORTS + max(RS.EMPTY_RESO= URCE - RESOURCE_STALLS.SCOREBOARD, 0)) / tma_info_thread_clks * (CYCLE_ACTI= VITY.STALLS_TOTAL - EXE_ACTIVITY.BOUND_ON_LOADS) / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -2084,7 +2083,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -2094,8 +2093,8 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= int_vector_128b, tma_int_vector_256b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6", "ScaleUnit": "100%" }, { @@ -2104,32 +2103,32 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote cache in other socket= s including synchronizations issues", - "MetricExpr": "((170 * tma_info_system_core_frequency - 37 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + (170 * = tma_info_system_core_frequency - 37 * tma_info_system_core_frequency) * MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD= _RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(133 * tma_info_system_core_frequency * MEM_LOAD_L3= _MISS_RETIRED.REMOTE_HITM + 133 * tma_info_system_core_frequency * MEM_LOAD= _L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETI= RED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote memory", - "MetricExpr": "(190 * tma_info_system_core_frequency - 37 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "153 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -2142,7 +2141,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks += tma_c02_wait", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -2151,8 +2150,8 @@ "MetricExpr": "tma_light_operations * INT_VEC_RETIRED.SHUFFLES / (= tma_retiring * tma_info_thread_slots)", "MetricGroup": "HPC;Pipeline;TopdownL4;tma_L4_group;tma_other_ligh= t_ops_group", "MetricName": "tma_shuffles_256b", - "MetricThreshold": "tma_shuffles_256b > 0.1 & tma_other_light_ops = > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers", + "MetricThreshold": "tma_shuffles_256b > 0.1 & (tma_other_light_ops= > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring Shuffle operations of 256-bit vector size (FP or In= teger). Shuffles may incur slow cross \"vector lane\" data transfers.", "ScaleUnit": "100%" }, { @@ -2161,7 +2160,7 @@ "MetricExpr": "CPU_CLK_UNHALTED.PAUSE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: CPU_CLK_UNHALTED.= PAUSE_INST", "ScaleUnit": "100%" }, @@ -2171,7 +2170,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -2179,8 +2178,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -2188,7 +2187,7 @@ "MetricExpr": "(XQ.FULL_CYCLES + L1D_PEND_MISS.L2_STALLS) / tma_in= fo_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -2197,8 +2196,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -2206,8 +2205,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -2215,8 +2214,8 @@ "MetricExpr": "(MEM_STORE_RETIRED.L2_HIT * 10 * (1 - MEM_INST_RETI= RED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_= LOADS / MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE= _REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -2233,7 +2232,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -2241,31 +2240,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2273,7 +2272,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -2282,7 +2281,7 @@ "MetricExpr": "INT_MISC.UNKNOWN_BRANCH_CYCLES / tma_info_thread_cl= ks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: FRONTEND_RETIRED.UNKNOWN_BRANCH", "ScaleUnit": "100%" }, @@ -2291,8 +2290,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 8639817A305 for ; Wed, 26 Mar 2025 00:09:19 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947763; cv=none; b=d9hUey5gH8ELWKw9A+0WJ2CyiVdL8cVlvfTwSXcNvu5b18JtazUl3mZC8bY0oiUv1jGN/O/zwO9ElftoGcivN+Bql3qxuLnc3xDkOKMulfLNwtJwdbCiG8JHijeSrbR/bgM36Wn7hRoA+nuK37ht4ED8p9Sm2UWj+940ThQgWQo= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947763; c=relaxed/simple; bh=mCT/VAKqn+UBHhr3FBRq3nJ06ddPW3FlPcpeak1SptM=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=eL939tD5tbqfS+V5mjOf8MBVgTWDhv9/VopiSFeu6rF5rQDbliSsxdN7nNdaUpurD7WhcYupf3W9tx6MtQs/oWcM+Mr6Sc+CdXRDNBuAw3F7MvzwIfi2J+GMq2eUZfeawj5cnePge9uYRUpOuIk53I87XeMlh+71tmj0A2DZgnw= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=rUMYAvTF; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="rUMYAvTF" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-301bbe9e084so17237464a91.0 for ; Tue, 25 Mar 2025 17:09:19 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947759; x=1743552559; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=LJCugyJkOrTEDqfvQOew0Lb9OgOoy9Dn0WMUEzjPQVU=; b=rUMYAvTFAEe7/vnT6l0xLqRdj4SR3SJ77S/792c7m5K7YMCFH//GiMbMQYUOKVjvsG FqKqPDyNwjWBexhpccuQgS9m6PHwPP/XKdiMmtBvE7yNHMyH9UeCuQCulw16muFQtGFG 3hX6bQLbtyDZG1XDUVaa8QUllJcVRINz16ivzEBdQWIOPmsvwXgh00Tfd6vlHZuomWmy vqIoDLFiAINKZuj7aA7eteEu7lTtnSSEc2YyljZChf8Yxw5lAV8IS5estWhapOA1ilX5 bssNdB4Ho7KC9/W1nItzMOxSD8iiYzh12fnQ97Fl4x9p4AA25ZGceETthSNhKFmCXySu jVtg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947759; x=1743552559; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=LJCugyJkOrTEDqfvQOew0Lb9OgOoy9Dn0WMUEzjPQVU=; b=WQgvFGA3tNy/ruly5+i0GhA8Vt7sJDru4STRh855NVYZ0zOx9/0yG+BCCq5wPaY7uK yGzwm8qdBreuk8Sxaiy05bKzOcJxMGhomS4HtkyhFKe0iHcPgFw0zE1z20Y5RddnWP4F rsdarcDXfhicWwoIjvhwCMs8WohLzpAZNN6fipjiY/QJEO0TDya8GJS9R5FtcsIOhRS3 cDTWYTgax4urmQwB6Lol1pVa0NhP/zunafi3VLK4FiYVMsiNZ2xQ4BEXttbYiGy0S7ga nRHJU5z9mm4h6pfbzAsocLBxx4h04hfqj2uAiGttIgj16L8/lyENIJD6sPgEPdq2x3ki nOKA== X-Forwarded-Encrypted: i=1; AJvYcCUu6pjzek08506FXqj11UtBu+0WHmOYlm3dlfx9oDnIGq0GDUA7iNxu9sZMWWTN2N9zKR3aEAHqaf11h+E=@vger.kernel.org X-Gm-Message-State: AOJu0YzBBaqfn+5ii/vYt0GUvK1Ol0q6+3OPySNqXW4uid3l1i7VMAiP x9uB0yIpGxt0wIQgG6q7acyaraHAztVeFNKsvLRI+ouCsWi0N1RMqcmvxzqe7O1UHGXdQVU1RFD MPx+LKQ== X-Google-Smtp-Source: AGHT+IG3q26+45eGfYtLnS4QbcOfqKK1CS3jAK1yXyjZ8X9V0NI3tjLL9H5uRLmupcZkHfwMAr27lLr5OnXT X-Received: from pjur8.prod.google.com ([2002:a17:90a:d408:b0:2fa:b84:b308]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:2688:b0:2ee:8ea0:6b9c with SMTP id 98e67ed59e1d1-3030fe90e28mr34606547a91.12.1742947759079; Tue, 25 Mar 2025 17:09:19 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:26 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-29-irogers@google.com> Subject: [PATCH v3 28/35] perf vendor events: Update sierraforest events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update events from v1.08 to v1.09. Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. The use of the spreadsheet for conversion has added thresholds. Signed-off-by: Ian Rogers --- tools/perf/pmu-events/arch/x86/mapfile.csv | 2 +- .../arch/x86/sierraforest/cache.json | 20 ++ .../arch/x86/sierraforest/memory.json | 20 ++ .../arch/x86/sierraforest/other.json | 48 ---- .../arch/x86/sierraforest/pipeline.json | 8 + .../arch/x86/sierraforest/srf-metrics.json | 204 ++++++++------- .../arch/x86/sierraforest/uncore-cache.json | 32 +++ .../arch/x86/sierraforest/uncore-memory.json | 240 ++++++++++++++++++ 8 files changed, 438 insertions(+), 136 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/mapfile.csv b/tools/perf/pmu-ev= ents/arch/x86/mapfile.csv index 0c16c9b840a5..bde2f32423a1 100644 --- a/tools/perf/pmu-events/arch/x86/mapfile.csv +++ b/tools/perf/pmu-events/arch/x86/mapfile.csv @@ -29,7 +29,7 @@ GenuineIntel-6-2E,v4,nehalemex,core GenuineIntel-6-A7,v1.04,rocketlake,core GenuineIntel-6-2A,v19,sandybridge,core GenuineIntel-6-8F,v1.25,sapphirerapids,core -GenuineIntel-6-AF,v1.08,sierraforest,core +GenuineIntel-6-AF,v1.09,sierraforest,core GenuineIntel-6-(37|4A|4C|4D|5A),v15,silvermont,core GenuineIntel-6-(4E|5E|8E|9E|A5|A6),v59,skylake,core GenuineIntel-6-55-[01234],v1.36,skylakex,core diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/cache.json b/tools= /perf/pmu-events/arch/x86/sierraforest/cache.json index 072df00aff92..21671c65d6dd 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/cache.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/cache.json @@ -466,6 +466,16 @@ "SampleAfterValue": "1000003", "UMask": "0x6" }, + { + "BriefDescription": "Counts demand data reads that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand data reads that were supplied b= y the L3 cache where a snoop was sent, the snoop hit, and modified data was= forwarded.", "Counter": "0,1,2,3,4,5,6,7", @@ -486,6 +496,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache where a snoop was sent, the snoop hit, and modified data was= forwarded.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/memory.json b/tool= s/perf/pmu-events/arch/x86/sierraforest/memory.json index 22d23077618e..3e2d0b565cfa 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/memory.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/memory.json @@ -82,6 +82,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0xB7", + "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x730000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/other.json b/tools= /perf/pmu-events/arch/x86/sierraforest/other.json index 4c77dac8ec78..daa16030d493 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/other.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/other.json @@ -8,46 +8,6 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, - { - "BriefDescription": "Counts demand data reads that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to this socket.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand data reads that were supplied b= y DRAM attached to another socket.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_DATA_RD.REMOTE_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x730000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0xB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3,4,5,6,7", @@ -57,13 +17,5 @@ "MSRValue": "0x10800", "SampleAfterValue": "100003", "UMask": "0x1" - }, - { - "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", - "Counter": "0,1,2,3,4,5,6,7", - "EventCode": "0x75", - "EventName": "SERIALIZATION.C01_MS_SCB", - "SampleAfterValue": "200003", - "UMask": "0x4" } ] diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/pipeline.json b/to= ols/perf/pmu-events/arch/x86/sierraforest/pipeline.json index df2c7bb474a0..a934b64f66d0 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/pipeline.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/pipeline.json @@ -300,6 +300,14 @@ "SampleAfterValue": "1000003", "UMask": "0x1" }, + { + "BriefDescription": "Counts the number of issue slots in a UMWAIT = or TPAUSE instruction where no uop issues due to the instruction putting th= e CPU into the C0.1 activity state.", + "Counter": "0,1,2,3,4,5,6,7", + "EventCode": "0x75", + "EventName": "SERIALIZATION.C01_MS_SCB", + "SampleAfterValue": "200003", + "UMask": "0x4" + }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend because allocation is stalled due to a mispredict= ed jump or a machine clear. [This event is alias to TOPDOWN_BAD_SPECULATION= .ALL_P]", "Counter": "0,1,2,3,4,5,6,7", diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/srf-metrics.json b= /tools/perf/pmu-events/arch/x86/sierraforest/srf-metrics.json index 83c86afd2960..ef629e4e91ce 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/srf-metrics.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/srf-metrics.json @@ -288,15 +288,17 @@ { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to certain allocation restrictions", "MetricExpr": "tma_core_bound", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_core_bound_group", + "MetricGroup": "TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_allocation_restriction", + "MetricThreshold": "tma_allocation_restriction > 0.1 & (tma_core_b= ound > 0.1 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend due to backend stalls", "MetricExpr": "TOPDOWN_BE_BOUND.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", + "MetricGroup": "TopdownL1;tma_L1_group", "MetricName": "tma_backend_bound", + "MetricThreshold": "tma_backend_bound > 0.1", "MetricgroupNoGroup": "TopdownL1", "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend due to backend stalls. Note that uops must= be available for consumption in order for this event to count. If a uop is= not available (IQ is empty), this event will not count", "ScaleUnit": "100%" @@ -304,92 +306,104 @@ { "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend because allocation is stalled due to a misp= redicted jump or a machine clear", "MetricExpr": "TOPDOWN_BAD_SPECULATION.ALL_P / (6 * CPU_CLK_UNHALT= ED.CORE)", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", + "MetricGroup": "TopdownL1;tma_L1_group", "MetricName": "tma_bad_speculation", + "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear", + "PublicDescription": "Counts the total number of issue slots that = were not consumed by the backend because allocation is stalled due to a mis= predicted jump or a machine clear. Only issue slots wasted due to fast nuke= s such as memory ordering nukes are counted. Other nukes are not accounted = for. Counts all issue slots blocked during this recovery window including r= elevant microcode flows and while uops are not yet available in the instruc= tion queue (IQ). Also includes the issue slots that were consumed by the ba= ckend but were thrown away because they were younger than the mispredict or= machine clear.", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to BACLEARS, which occurs when the Branch T= arget Buffer (BTB) prediction or lack thereof, was corrected by a later bra= nch predictor in the frontend", "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_DETECT / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_detect", - "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s", + "MetricThreshold": "tma_branch_detect > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", + "PublicDescription": "Counts the number of issue slots that were n= ot delivered by the frontend due to BACLEARS, which occurs when the Branch = Target Buffer (BTB) prediction or lack thereof, was corrected by a later br= anch predictor in the frontend. Includes BACLEARS due to all branch types i= ncluding conditional and unconditional jumps, returns, and indirect branche= s.", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to branch mispredicts", "MetricExpr": "TOPDOWN_BAD_SPECULATION.MISPREDICT / (6 * CPU_CLK_U= NHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_bad_speculation_g= roup", + "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_branch_mispredicts", + "MetricThreshold": "tma_branch_mispredicts > 0.05 & tma_bad_specul= ation > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to BTCLEARS, which occurs when the Branch T= arget Buffer (BTB) predicts a taken branch", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to BTCLEARS, which occurs when the Branch T= arget Buffer (BTB) predicts a taken branch.", "MetricExpr": "TOPDOWN_FE_BOUND.BRANCH_RESTEER / (6 * CPU_CLK_UNHA= LTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_branch_resteer", + "MetricThreshold": "tma_branch_resteer > 0.05 & (tma_ifetch_latenc= y > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to the microcode sequencer (MS)", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to the microcode sequencer (MS).", "MetricExpr": "TOPDOWN_FE_BOUND.CISC / (6 * CPU_CLK_UNHALTED.CORE)= ", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_cisc", + "MetricThreshold": "tma_cisc > 0.05 & (tma_ifetch_bandwidth > 0.1 = & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of cycles due to backend bo= und stalls that are bounded by core restrictions and not attributed to an o= utstanding load or stores, or resource limitation", "MetricExpr": "TOPDOWN_BE_BOUND.ALLOC_RESTRICTIONS / (6 * CPU_CLK_= UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_backend_bound_gro= up", + "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_core_bound", + "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.1= ", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to decode stalls", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to decode stalls.", "MetricExpr": "TOPDOWN_FE_BOUND.DECODE / (6 * CPU_CLK_UNHALTED.COR= E)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_decode", + "MetricThreshold": "tma_decode > 0.05 & (tma_ifetch_bandwidth > 0.= 1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to a machine clear that does not require the = use of microcode, classified as a fast nuke, due to memory ordering, memory= disambiguation and memory renaming", "MetricExpr": "TOPDOWN_BAD_SPECULATION.FASTNUKE / (6 * CPU_CLK_UNH= ALTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_machine_clears_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_fast_nuke", + "MetricThreshold": "tma_fast_nuke > 0.05 & (tma_machine_clears > 0= .05 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to frontend stalls", + "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to frontend stalls.", "MetricExpr": "TOPDOWN_FE_BOUND.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", + "MetricGroup": "TopdownL1;tma_L1_group", "MetricName": "tma_frontend_bound", + "MetricThreshold": "tma_frontend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to instruction cache misses", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to instruction cache misses.", "MetricExpr": "TOPDOWN_FE_BOUND.ICACHE / (6 * CPU_CLK_UNHALTED.COR= E)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_icache_misses", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_ifetch_latency= > 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend bandwidth restrictions due to d= ecode, predecode, cisc, and other limitations", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend bandwidth restrictions due to d= ecode, predecode, cisc, and other limitations.", "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_BANDWIDTH / (6 * CPU_CLK_= UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_frontend_bound_gr= oup", + "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_bandwidth", + "MetricThreshold": "tma_ifetch_bandwidth > 0.1 & tma_frontend_boun= d > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend latency restrictions due to ica= che misses, itlb misses, branch detection, and resteer limitations", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to frontend latency restrictions due to ica= che misses, itlb misses, branch detection, and resteer limitations.", "MetricExpr": "TOPDOWN_FE_BOUND.FRONTEND_LATENCY / (6 * CPU_CLK_UN= HALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_frontend_bound_gr= oup", + "MetricGroup": "TopdownL2;tma_L2_group;tma_frontend_bound_group", "MetricName": "tma_ifetch_latency", + "MetricThreshold": "tma_ifetch_latency > 0.15 & tma_frontend_bound= > 0.2", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, @@ -420,33 +434,28 @@ { "BriefDescription": "Percentage of time that retirement is stalled= due to a first level data TLB miss", "MetricExpr": "100 * (LD_HEAD.DTLB_MISS_AT_RET + LD_HEAD.PGWALK_AT= _RET) / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "Cycles", - "MetricName": "tma_info_bottleneck_dtlb_miss_bound_cycles", - "ScaleUnit": "100%" + "MetricName": "tma_info_bottleneck_%_dtlb_miss_bound_cycles" }, { "BriefDescription": "Percentage of time that allocation and retire= ment is stalled by the Frontend Cluster due to an Ifetch Miss, either Icach= e or ITLB Miss", "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.ALL / CPU_CLK_UNHALTE= D.CORE", - "MetricGroup": "Cycles;Ifetch", - "MetricName": "tma_info_bottleneck_ifetch_miss_bound_cycles", - "PublicDescription": "Percentage of time that allocation and retir= ement is stalled by the Frontend Cluster due to an Ifetch Miss, either Icac= he or ITLB Miss. See Info.Ifetch_Bound", - "ScaleUnit": "100%" + "MetricGroup": "Ifetch", + "MetricName": "tma_info_bottleneck_%_ifetch_miss_bound_cycles", + "PublicDescription": "Percentage of time that allocation and retir= ement is stalled by the Frontend Cluster due to an Ifetch Miss, either Icac= he or ITLB Miss. See Info.Ifetch_Bound" }, { "BriefDescription": "Percentage of time that retirement is stalled= due to an L1 miss", "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.ALL / CPU_CLK_UNHALTED.= CORE", - "MetricGroup": "Cycles;Load_Store_Miss", - "MetricName": "tma_info_bottleneck_load_miss_bound_cycles", - "PublicDescription": "Percentage of time that retirement is stalle= d due to an L1 miss. See Info.Load_Miss_Bound", - "ScaleUnit": "100%" + "MetricGroup": "Load_Store_Miss", + "MetricName": "tma_info_bottleneck_%_load_miss_bound_cycles", + "PublicDescription": "Percentage of time that retirement is stalle= d due to an L1 miss. See Info.Load_Miss_Bound" }, { "BriefDescription": "Percentage of time that retirement is stalled= by the Memory Cluster due to a pipeline stall", "MetricExpr": "100 * LD_HEAD.ANY_AT_RET / CPU_CLK_UNHALTED.CORE", - "MetricGroup": "Cycles;Mem_Exec", - "MetricName": "tma_info_bottleneck_mem_exec_bound_cycles", - "PublicDescription": "Percentage of time that retirement is stalle= d by the Memory Cluster due to a pipeline stall. See Info.Mem_Exec_Bound", - "ScaleUnit": "100%" + "MetricGroup": "Mem_Exec", + "MetricName": "tma_info_bottleneck_%_mem_exec_bound_cycles", + "PublicDescription": "Percentage of time that retirement is stalle= d by the Memory Cluster due to a pipeline stall. See Info.Mem_Exec_Bound" }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -501,26 +510,22 @@ { "BriefDescription": "Percentage of time that allocation is stalled= due to load buffer full", "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.LD_BUF / CPU_CLK_UNHALTED= .CORE", - "MetricName": "tma_info_buffer_stalls_load_buffer_stall_cycles", - "ScaleUnit": "100%" + "MetricName": "tma_info_buffer_stalls_%_load_buffer_stall_cycles" }, { "BriefDescription": "Percentage of time that allocation is stalled= due to memory reservation stations full", "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.RSV / CPU_CLK_UNHALTED.CO= RE", - "MetricName": "tma_info_buffer_stalls_mem_rsv_stall_cycles", - "ScaleUnit": "100%" + "MetricName": "tma_info_buffer_stalls_%_mem_rsv_stall_cycles" }, { "BriefDescription": "Percentage of time that allocation is stalled= due to store buffer full", "MetricExpr": "100 * MEM_SCHEDULER_BLOCK.ST_BUF / CPU_CLK_UNHALTED= .CORE", - "MetricName": "tma_info_buffer_stalls_store_buffer_stall_cycles", - "ScaleUnit": "100%" + "MetricName": "tma_info_buffer_stalls_%_store_buffer_stall_cycles" }, { "BriefDescription": "Cycles Per Instruction", "MetricExpr": "CPU_CLK_UNHALTED.CORE / INST_RETIRED.ANY", - "MetricName": "tma_info_core_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_core_cpi" }, { "BriefDescription": "Floating Point Operations Per Cycle", @@ -541,28 +546,46 @@ { "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss hits in the L2", "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.L2_HIT / MEM_BOUND_ST= ALLS_IFETCH.ALL", - "MetricName": "tma_info_ifetch_miss_bound_ifetchmissbound_with_l2h= it", - "ScaleUnit": "100%" + "MetricName": "tma_info_ifetch_miss_bound_%_ifetchmissbound_with_l= 2hit" + }, + { + "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss doesn't hit in the L2", + "MetricExpr": "100 * (MEM_BOUND_STALLS_IFETCH.LLC_HIT + MEM_BOUND_= STALLS_IFETCH.LLC_MISS) / MEM_BOUND_STALLS_IFETCH.ALL", + "MetricName": "tma_info_ifetch_miss_bound_%_ifetchmissbound_with_l= 2miss" }, { "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss hits in the L3", "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.LLC_HIT / MEM_BOUND_S= TALLS_IFETCH.ALL", - "MetricName": "tma_info_ifetch_miss_bound_ifetchmissbound_with_l3h= it", - "ScaleUnit": "100%" + "MetricName": "tma_info_ifetch_miss_bound_%_ifetchmissbound_with_l= 3hit" + }, + { + "BriefDescription": "Percentage of ifetch miss bound stalls, where= the ifetch miss subsequently misses in the L3", + "MetricExpr": "100 * MEM_BOUND_STALLS_IFETCH.LLC_MISS / MEM_BOUND_= STALLS_IFETCH.ALL", + "MetricName": "tma_info_ifetch_miss_bound_%_ifetchmissbound_with_l= 3miss" }, { "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that hit the L2", "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.L2_HIT / MEM_BOUND_STAL= LS_LOAD.ALL", "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_miss_bound_loadmissbound_with_l2hit", - "ScaleUnit": "100%" + "MetricName": "tma_info_load_miss_bound_%_loadmissbound_with_l2hit" + }, + { + "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that subsequently misses in the L2", + "MetricExpr": "100 * (MEM_BOUND_STALLS_LOAD.LLC_HIT + MEM_BOUND_ST= ALLS_LOAD.LLC_MISS) / MEM_BOUND_STALLS_LOAD.ALL", + "MetricGroup": "load_store_bound", + "MetricName": "tma_info_load_miss_bound_%_loadmissbound_with_l2mis= s" }, { "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that hit the L3", "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.LLC_HIT / MEM_BOUND_STA= LLS_LOAD.ALL", "MetricGroup": "load_store_bound", - "MetricName": "tma_info_load_miss_bound_loadmissbound_with_l3hit", - "ScaleUnit": "100%" + "MetricName": "tma_info_load_miss_bound_%_loadmissbound_with_l3hit" + }, + { + "BriefDescription": "Percentage of memory bound stalls where retir= ement is stalled due to an L1 miss that subsequently misses the L3", + "MetricExpr": "100 * MEM_BOUND_STALLS_LOAD.LLC_MISS / MEM_BOUND_ST= ALLS_LOAD.ALL", + "MetricGroup": "load_store_bound", + "MetricName": "tma_info_load_miss_bound_%_loadmissbound_with_l3mis= s" }, { "BriefDescription": "Counts the number of cycles that the oldest l= oad of the load buffer is stalled at retirement due to a pipeline block", @@ -600,44 +623,37 @@ { "BriefDescription": "Percentage of total non-speculative loads wit= h an address aliasing block", "MetricExpr": "100 * LD_BLOCKS.ADDRESS_ALIAS / MEM_UOPS_RETIRED.AL= L_LOADS", - "MetricName": "tma_info_mem_exec_blocks_loads_with_adressaliasing", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_blocks_%_loads_with_adressaliasin= g" }, { "BriefDescription": "Percentage of total non-speculative loads wit= h a store forward or unknown store address block", "MetricExpr": "100 * LD_BLOCKS.DATA_UNKNOWN / MEM_UOPS_RETIRED.ALL= _LOADS", - "MetricName": "tma_info_mem_exec_blocks_loads_with_storefwdblk", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_blocks_%_loads_with_storefwdblk" }, { "BriefDescription": "Percentage of Memory Execution Bound due to a= first level data cache miss", "MetricExpr": "100 * LD_HEAD.L1_MISS_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_l1miss", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_bound_%_loadhead_with_l1miss" }, { "BriefDescription": "Percentage of Memory Execution Bound due to o= ther block cases, such as pipeline conflicts, fences, etc", "MetricExpr": "100 * LD_HEAD.OTHER_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_otherpipeline= blks", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_bound_%_loadhead_with_otherpipeli= neblks" }, { "BriefDescription": "Percentage of Memory Execution Bound due to a= pagewalk", "MetricExpr": "100 * LD_HEAD.PGWALK_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_pagewalk", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_bound_%_loadhead_with_pagewalk" }, { "BriefDescription": "Percentage of Memory Execution Bound due to a= second level TLB miss", "MetricExpr": "100 * LD_HEAD.DTLB_MISS_AT_RET / LD_HEAD.ANY_AT_RET= ", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_stlbhit", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_bound_%_loadhead_with_stlbhit" }, { "BriefDescription": "Percentage of Memory Execution Bound due to a= store forward address match", "MetricExpr": "100 * LD_HEAD.ST_ADDR_AT_RET / LD_HEAD.ANY_AT_RET", - "MetricName": "tma_info_mem_exec_bound_loadhead_with_storefwding", - "ScaleUnit": "100%" + "MetricName": "tma_info_mem_exec_bound_%_loadhead_with_storefwding" }, { "BriefDescription": "Instructions per Load", @@ -667,8 +683,7 @@ { "BriefDescription": "Percentage of time that the core is stalled d= ue to a TPAUSE or UMWAIT instruction", "MetricExpr": "100 * SERIALIZATION.C01_MS_SCB / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricName": "tma_info_serialization_tpause_cycles", - "ScaleUnit": "100%" + "MetricName": "tma_info_serialization_%_tpause_cycles" }, { "BriefDescription": "Average CPU Utilization", @@ -684,17 +699,20 @@ }, { "BriefDescription": "Fraction of cycles spent in Kernel mode", - "MetricExpr": "CPU_CLK_UNHALTED.CORE_P:k / CPU_CLK_UNHALTED.CORE", + "MetricExpr": "cpu@CPU_CLK_UNHALTED.CORE_P@k / CPU_CLK_UNHALTED.CO= RE", + "MetricGroup": "Summary", "MetricName": "tma_info_system_kernel_utilization" }, { "BriefDescription": "PerfMon Event Multiplexing accuracy indicator= ", "MetricExpr": "CPU_CLK_UNHALTED.CORE_P / CPU_CLK_UNHALTED.CORE", - "MetricName": "tma_info_system_mux" + "MetricName": "tma_info_system_mux", + "MetricThreshold": "tma_info_system_mux > 1.1 | tma_info_system_mu= x < 0.9" }, { "BriefDescription": "Average Frequency Utilization relative nomina= l frequency", "MetricExpr": "CPU_CLK_UNHALTED.CORE / CPU_CLK_UNHALTED.REF_TSC", + "MetricGroup": "Power", "MetricName": "tma_info_system_turbo_utilization" }, { @@ -718,90 +736,102 @@ "MetricName": "tma_info_uop_mix_x87_uop_ratio" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to Instruction Table Lookaside Buffer (ITLB= ) misses", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to Instruction Table Lookaside Buffer (ITLB= ) misses.", "MetricExpr": "TOPDOWN_FE_BOUND.ITLB_MISS / (6 * CPU_CLK_UNHALTED.= CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_latency_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_latency_group", "MetricName": "tma_itlb_misses", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_ifetch_latency >= 0.15 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the total number of issue slots that w= ere not consumed by the backend because allocation is stalled due to a mach= ine clear (nuke) of any kind including memory ordering and memory disambigu= ation", "MetricExpr": "TOPDOWN_BAD_SPECULATION.MACHINE_CLEARS / (6 * CPU_C= LK_UNHALTED.CORE)", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_bad_speculation_g= roup", + "MetricGroup": "TopdownL2;tma_L2_group;tma_bad_speculation_group", "MetricName": "tma_machine_clears", + "MetricThreshold": "tma_machine_clears > 0.05 & tma_bad_speculatio= n > 0.15", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to memory reservation stalls in which a sched= uler is not able to accept uops", "MetricExpr": "TOPDOWN_BE_BOUND.MEM_SCHEDULER / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_mem_scheduler", + "MetricThreshold": "tma_mem_scheduler > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to IEC or FPC RAT stalls, which can be due to= FIQ or IEC reservation stalls in which the integer, floating point or SIMD= scheduler is not able to accept uops", "MetricExpr": "TOPDOWN_BE_BOUND.NON_MEM_SCHEDULER / (6 * CPU_CLK_U= NHALTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_non_mem_scheduler", + "MetricThreshold": "tma_non_mem_scheduler > 0.1 & (tma_resource_bo= und > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to a machine clear that requires the use of m= icrocode (slow nuke)", "MetricExpr": "TOPDOWN_BAD_SPECULATION.NUKE / (6 * CPU_CLK_UNHALTE= D.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_machine_clears_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_machine_clears_group", "MetricName": "tma_nuke", + "MetricThreshold": "tma_nuke > 0.05 & (tma_machine_clears > 0.05 &= tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to other common frontend stalls not categor= ized", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to other common frontend stalls not categor= ized.", "MetricExpr": "TOPDOWN_FE_BOUND.OTHER / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_other_fb", + "MetricThreshold": "tma_other_fb > 0.05 & (tma_ifetch_bandwidth > = 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to wrong predecodes", + "BriefDescription": "Counts the number of issue slots that were no= t delivered by the frontend due to wrong predecodes.", "MetricExpr": "TOPDOWN_FE_BOUND.PREDECODE / (6 * CPU_CLK_UNHALTED.= CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_ifetch_bandwidth_= group", + "MetricGroup": "TopdownL3;tma_L3_group;tma_ifetch_bandwidth_group", "MetricName": "tma_predecode", + "MetricThreshold": "tma_predecode > 0.05 & (tma_ifetch_bandwidth >= 0.1 & tma_frontend_bound > 0.2)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to the physical register file unable to accep= t an entry (marble stalls)", "MetricExpr": "TOPDOWN_BE_BOUND.REGISTER / (6 * CPU_CLK_UNHALTED.C= ORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_register", + "MetricThreshold": "tma_register > 0.1 & (tma_resource_bound > 0.2= & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to the reorder buffer being full (ROB stalls)= ", "MetricExpr": "TOPDOWN_BE_BOUND.REORDER_BUFFER / (6 * CPU_CLK_UNHA= LTED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_reorder_buffer", + "MetricThreshold": "tma_reorder_buffer > 0.1 & (tma_resource_bound= > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of cycles the core is stall= ed due to a resource limitation", "MetricExpr": "tma_backend_bound - tma_core_bound", - "MetricGroup": "Slots;TopdownL2;tma_L2_group;tma_backend_bound_gro= up", + "MetricGroup": "TopdownL2;tma_L2_group;tma_backend_bound_group", "MetricName": "tma_resource_bound", + "MetricThreshold": "tma_resource_bound > 0.2 & tma_backend_bound >= 0.1", "MetricgroupNoGroup": "TopdownL2", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that result = in retirement slots", "MetricExpr": "TOPDOWN_RETIRING.ALL_P / (6 * CPU_CLK_UNHALTED.CORE= )", - "MetricGroup": "Slots;TopdownL1;tma_L1_group", + "MetricGroup": "TopdownL1;tma_L1_group", "MetricName": "tma_retiring", + "MetricThreshold": "tma_retiring > 0.75", "MetricgroupNoGroup": "TopdownL1", "ScaleUnit": "100%" }, { "BriefDescription": "Counts the number of issue slots that were no= t consumed by the backend due to scoreboards from the instruction queue (IQ= ), jump execution unit (JEU), or microcode sequencer (MS)", "MetricExpr": "TOPDOWN_BE_BOUND.SERIALIZATION / (6 * CPU_CLK_UNHAL= TED.CORE)", - "MetricGroup": "Slots;TopdownL3;tma_L3_group;tma_resource_bound_gr= oup", + "MetricGroup": "TopdownL3;tma_L3_group;tma_resource_bound_group", "MetricName": "tma_serialization", + "MetricThreshold": "tma_serialization > 0.1 & (tma_resource_bound = > 0.2 & tma_backend_bound > 0.1)", "ScaleUnit": "100%" }, { diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/uncore-cache.json = b/tools/perf/pmu-events/arch/x86/sierraforest/uncore-cache.json index a779a1a73ea5..7182ca00ef8d 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/uncore-cache.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/uncore-cache.json @@ -873,6 +873,38 @@ "UMask": "0x1", "Unit": "CHA" }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking exclusive lines in the cores? cache.? Snoop filter capacity = evictions occur when the snoop filter is full and evicts an existing entry = to track a new entry.? Does not count clean evictions such as when a core?s= cache replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.E_STATE", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : E state", + "UMask": "0x2", + "Unit": "CHA" + }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking modified lines in the cores? cache.? Snoop filter capacity e= victions occur when the snoop filter is full and evicts an existing entry t= o track a new entry.? Does not count clean evictions such as when a core?s = cache replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.M_STATE", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : M state", + "UMask": "0x1", + "Unit": "CHA" + }, + { + "BriefDescription": "Counts snoop filter capacity evictions for en= tries tracking shared lines in the cores? cache.? Snoop filter capacity evi= ctions occur when the snoop filter is full and evicts an existing entry to = track a new entry.? Does not count clean evictions such as when a core?s ca= che replaces a tracked cacheline with a new cacheline.", + "Counter": "0,1,2,3", + "EventCode": "0x3d", + "EventName": "UNC_CHA_SF_EVICTION.S_STATE", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "Snoop Filter Capacity Evictions : S state", + "UMask": "0x4", + "Unit": "CHA" + }, { "BriefDescription": "All TOR Inserts", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/sierraforest/uncore-memory.json= b/tools/perf/pmu-events/arch/x86/sierraforest/uncore-memory.json index ae9c62b32e92..c7e9dbe02eb0 100644 --- a/tools/perf/pmu-events/arch/x86/sierraforest/uncore-memory.json +++ b/tools/perf/pmu-events/arch/x86/sierraforest/uncore-memory.json @@ -188,6 +188,94 @@ "PublicDescription": "DRAM Clockticks", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 temp readings forced 2x refre= sh", + "Counter": "0,1,2,3", + "EventCode": "0xA7", + "EventName": "UNC_M_MR4_2XREF_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH0_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles MR4 MRRs was triggered/running", + "Counter": "0,1,2,3", + "EventCode": "0xA6", + "EventName": "UNC_M_PDC_MR4ACTIVE_CYCLES.SCH1_DIMM1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, { "BriefDescription": "# of cycles a given rank is in Power Down Mod= e", "Counter": "0,1,2,3", @@ -286,6 +374,70 @@ "PublicDescription": "-", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM and throttle level is zero.", + "Counter": "0,1,2,3", + "EventCode": "0x89", + "EventName": "UNC_M_POWER_CRITICAL_THROTTLE_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT0", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.BW_SLOT1", + "Experimental": "1", + "PerPkg": "1", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "MR4 temp reading is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.MR4BLKEN", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x8", + "Unit": "IMC" + }, + { + "BriefDescription": "RAPL is throttling", + "Counter": "0,1,2,3", + "EventCode": "0x46", + "EventName": "UNC_M_POWER_THROTTLE_CYCLES.RAPLBLK", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x4", + "Unit": "IMC" + }, { "BriefDescription": "DRAM Precharge commands. : Counts the number = of DRAM Precharge commands sent on this channel.", "Counter": "0,1,2,3", @@ -480,6 +632,94 @@ "UMask": "0x1", "Unit": "IMC" }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Critical level on s= pecified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8e", + "EventName": "UNC_M_THROTTLE_CRIT_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at High level on speci= fied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8d", + "EventName": "UNC_M_THROTTLE_HIGH_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Normal level on spe= cified DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8b", + "EventName": "UNC_M_THROTTLE_LOW_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT0", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x1", + "Unit": "IMC" + }, + { + "BriefDescription": "# of cycles Throttling at Mid level on specif= ied DIMM", + "Counter": "0,1,2,3", + "EventCode": "0x8c", + "EventName": "UNC_M_THROTTLE_MID_CYCLES.SLOT1", + "Experimental": "1", + "PerPkg": "1", + "PublicDescription": "-", + "UMask": "0x2", + "Unit": "IMC" + }, { "BriefDescription": "Write Pending Queue Allocations", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 09B8E17B50A for ; Wed, 26 Mar 2025 00:09:29 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947776; cv=none; b=WF1YgMig3PfJipmk/fVUxCpe8mbygNkqWDgvKPqyg4HG7fFuEAIuwZivcuIG+KnyjmOhR+ivCxBTOYB51H9XV6YDehfWnqvTXV3CI7VTPASBVta+UuuBVr+qxnSSjtzZsxbINQrwamoeNlNsV80//tQenQwsAkf+ZtfGsUxFzqI= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947776; c=relaxed/simple; bh=yXsuB7eBiZQnUjcSPmlU08BWoINyiu6kQe45F7dclcY=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=k+wZ+20RltozQBFhFhgE5oABSSk1aaF/j+Ol507Dh6t8mKLgGUB8tLuhktCVJeE/Tb8d7SEUqTPYXryY2HaEHHXffevXj9MIM0/TBmPeLTkvndGluYj5mwYzOGBhkGIHyEEMYOwhv725NEPyCQMQCoiI5skDYOnWG/FxjTvOr2k= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=31G5gY2Z; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="31G5gY2Z" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff52e1c56fso16292484a91.2 for ; Tue, 25 Mar 2025 17:09:29 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947768; x=1743552568; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=gXuRrw3IxGYFDpF/XF089NtvMxpystzkyGvzqRqcA/0=; b=31G5gY2Zql4bpBsZKkFvXwVCcq1r7J2P5uV77dOSH4OPzQXHdlgFxAADR9unr/pSvG 9HlrbNIR+eRvtQXTkTq48MmxO09+UftyYZDFvMHUfpKNKWJojTCeIsi+sWFN+pioixw9 skadZBFTGE+hD8nng/m6TErjtO3JmZU8j2rjRvhx7OiFd0nW77qVi4StvOxbl4/1OzWd hMHcpmSHGGZXJxP2gPqw36YnaUEWYugsPC1SgahSj3fWxU+trj2dOoBwUtWGhsd3pysP 274pc/T6v9vhshb2Xp3G7Qr/ZlfQeU1Lo08oHZfU2uGPeh9RfBBVHNbA4m0b/CRJvHux ZjjQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947768; x=1743552568; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=gXuRrw3IxGYFDpF/XF089NtvMxpystzkyGvzqRqcA/0=; b=aFOoFuAtGgy77XyM+iv68Cl/8PZtIKmF1Yrop2vGRdMQcMbhglqtA1VICLYFeQAwY1 bCO1NeoZSsT9Ji4xh/ONPwo4PlWtz9DTCePqpDiB/DHjNRlyn6tZhctLv5hcrRlQUF5A YmEbUdyUJO9YSRzEgb+BczqfGnGc7EP/+cCBWBld2jule2L1AVhecaBPlYrt2Ydmbk4f g2Tpfy0ERN9tPTFUSiwminMyGnapsL3qYEKoB9zvIvgaSVGOfK/C0terv3JAXFdyYB2E CTrD5zhbfYTgiwRQ1VVEnKtDI93h7rw+HbNkX66y8X04AONxntY9ltfm1ATL0XpYNpFT rfOQ== X-Forwarded-Encrypted: i=1; AJvYcCVx0swQYMiGMgOF6QDGXWMS88cw3zYXSCl2P52u+AHOi3rqt7x9lauKaalrakCMJPeBDpVwnLJSDITzYkU=@vger.kernel.org X-Gm-Message-State: AOJu0YygJZw26UTw7QAs41B5Kf9qVHYxsZcKwYEXgW1++amJqNERyi+P 75WuVyOi3rn89Gws2uKd4M6X4nUpiRKAXxWys22f8+EhlV8KpxiZ0E224uQSB1HwHazke8MB+ye tykPwig== X-Google-Smtp-Source: AGHT+IGF69bPe3ydiQRBC0mGle5UA9UbcoxmaHyaAxGPDQyx6NVdQelGAcfx519/PZqI5AlhQZQF2R/7BOLz X-Received: from pgc16.prod.google.com ([2002:a05:6a02:2f90:b0:af2:220e:6804]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:168e:b0:1f5:709d:e0cb with SMTP id adf61e73a8af0-1fe43437231mr31611763637.39.1742947768336; Tue, 25 Mar 2025 17:09:28 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:27 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-30-irogers@google.com> Subject: [PATCH v3 29/35] perf vendor events: Update skylake metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/skylake/skl-metrics.json | 367 +++++++++--------- 1 file changed, 183 insertions(+), 184 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/skylake/skl-metrics.json b/tool= s/perf/pmu-events/arch/x86/skylake/skl-metrics.json index 2a76dd01fb52..2d3a037e88b5 100644 --- a/tools/perf/pmu-events/arch/x86/skylake/skl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/skylake/skl-metrics.json @@ -74,12 +74,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -91,7 +91,7 @@ "MetricExpr": "34 * (FP_ASSIST.ANY + OTHER_ASSISTS.ANY) / tma_info= _thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, @@ -102,7 +102,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -112,12 +112,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -132,7 +132,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -140,7 +140,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_dtlb_store)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound= + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_= store_latency / (tma_store_latency + tma_false_sharing + tma_split_stores += tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency)) + tma_memory_bound * (tma_store_bound / (tma_dram_bou= nd + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_= store_latency / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tm= a_store_latency)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -148,22 +148,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_b= ranches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tm= a_ms_switches + tma_lcp + tma_dsb_switches)) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_b= ranches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + t= ma_itlb_misses + tma_lcp + tma_ms_switches)) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_mispredicts_resteers = + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_i= tlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_clears_resteers + tma= _mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma= _dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -171,7 +171,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -179,15 +179,15 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_dtlb_store - tma_store_latency)) + tma_machine_clears * (1 - tma_o= ther_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency - tma_store_latency)) + tma_machine_clears * (1 - tma_o= ther_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears" + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -199,10 +199,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -224,8 +224,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -233,8 +233,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -242,7 +242,7 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, @@ -251,7 +251,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -259,33 +259,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((22 * tma_info_system_core_frequency - 3.5 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM + (20 * tma_= info_system_core_frequency - 3.5 * tma_info_system_core_frequency) * MEM_LO= AD_L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETI= RED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(18.5 * tma_info_system_core_frequency * MEM_LOAD_L= 3_HIT_RETIRED.XSNP_HITM + 16.5 * tma_info_system_core_frequency * MEM_LOAD_= L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED= .L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -296,25 +296,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(20 * tma_info_system_core_frequency - 3.5 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT * (1 + MEM_LOA= D_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "16.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _HIT_RETIRED.XSNP_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_= MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -323,7 +323,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -333,7 +333,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -343,7 +343,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -351,27 +351,27 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -380,18 +380,18 @@ "MetricExpr": "22 * tma_info_system_core_frequency * OFFCORE_RESPO= NSE.DEMAND_RFO.L3_HIT.SNOOP_HITM / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related m= etrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_= data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tm= a_data_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -401,7 +401,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -411,7 +411,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -431,7 +431,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -440,7 +440,7 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { @@ -448,8 +448,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports= _utilized_2", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -458,8 +458,8 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", + "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -467,8 +467,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -476,8 +476,8 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", - "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_port_0, tma_port_1, tma_port_5, tma_p= ort_6, tma_ports_utilized_2", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", + "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -487,35 +487,35 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / U= OPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "(UOPS_RETIRED.RETIRE_SLOTS + UOPS_RETIRED.MACRO_FUS= ED - INST_RETIRED.ANY) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", - "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D0x1\\,edge\\=3D0x1@) / tma_info_thread_clks", + "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D1\\,edge@) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -526,11 +526,11 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -555,7 +555,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= )))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= )))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -564,7 +564,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -572,10 +572,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -604,7 +605,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -632,11 +633,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -649,20 +650,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHE= S.COUNT", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@ + 2", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -698,7 +699,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -717,7 +718,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -725,7 +726,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -733,7 +734,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -741,7 +742,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -749,7 +750,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -799,7 +800,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -974,8 +975,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -996,12 +997,12 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ASSIST.ANY + OTHER_ASSISTS.A= NY)", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1043,14 +1044,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1061,7 +1061,7 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_ARB_TRK_OCCUPANCY.DATA_READ / UNC_ARB_TRK_OCCUP= ANCY.DATA_READ@cmask\\=3D0x1@", + "MetricExpr": "UNC_ARB_TRK_OCCUPANCY.DATA_READ / UNC_ARB_TRK_OCCUP= ANCY.DATA_READ@cmask\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1112,7 +1112,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1121,15 +1121,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1155,15 +1154,15 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1171,7 +1170,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1180,17 +1179,17 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates how often the CPU was s= talled due to L2 cache accesses by loads", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D0x1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2= _MISS) / tma_info_thread_clks)", + "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_M= ISS) / tma_info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1199,7 +1198,7 @@ "MetricExpr": "3.5 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1208,17 +1207,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(10 * tma_info_system_core_frequency - 3.5 * tma_in= fo_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETIRE= D.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "6.5 * tma_info_system_core_frequency * (MEM_LOAD_RE= TIRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)= ) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1226,18 +1225,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1255,7 +1254,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1263,31 +1262,31 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1295,7 +1294,7 @@ "MetricExpr": "(12 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (9 = * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAND= ING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1307,15 +1306,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1324,7 +1323,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1336,11 +1335,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1362,7 +1361,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1376,12 +1375,12 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { @@ -1389,7 +1388,7 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1399,7 +1398,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1407,8 +1406,8 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / UOPS_RETI= RED.RETIRE_SLOTS", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP_PS", "ScaleUnit": "100%" }, { @@ -1421,19 +1420,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1442,7 +1441,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_1, = tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1451,7 +1450,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_port_0, tma_port_5, tma_port_6, tma_ports_= utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1487,7 +1486,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_sc= alar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tm= a_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1496,7 +1495,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, t= ma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1513,8 +1512,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1522,8 +1521,8 @@ "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_cl= ks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1531,7 +1530,7 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CO= RE_CYCLES_GE_2) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1540,16 +1539,16 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CO= RE_CYCLES_GE_3) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_port_0, tma_port_1, tma_port_5, tma_port_6", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", "MetricExpr": "(UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else= UOPS_EXECUTED.CORE_CYCLES_GE_3) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1567,7 +1566,7 @@ "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clk= s", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related me= trics: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1578,7 +1577,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1586,8 +1585,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1595,7 +1594,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1604,8 +1603,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1613,8 +1612,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1623,8 +1622,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 9 * (1 - MEM_INST_RETIRED.LOCK_= LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / M= EM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS= _OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1640,7 +1639,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -1648,31 +1647,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1680,7 +1679,7 @@ "MetricExpr": "9 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1689,8 +1688,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 4DC8C188580 for ; Wed, 26 Mar 2025 00:09:38 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947786; cv=none; b=aqAXE8Iqpc6PL96GVhIEntlQhS0wxozOmOuUCr2O040FLGRoeUgV+EH7gCCMyW1zAhXDcHYB99Xww18FnWTnCX6AeC9+U3wKLv1hOay8Taf0aQCm3D0HI0UOtkQXu1Slmnnud0qA5QZ6CfH2Nal/J68OvRDaxd8HnMkNfz7NBgY= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947786; c=relaxed/simple; bh=Smd1jKiEWEGZJ6nNESE82L+COZ29iPn3G/A1ZhFwfeM=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=PiZootmTzomDRVFk9Z4MpC4d6cfycR6XDDesLLziZBuuVggxn2zEm9aQYNAX0edxGNN2S6v9PxvdvFe6q6CdSut6B6TarCqHOTQPdPIRJUxSzNW+pjAv+x6XjwD8E5FXuj2hT/ghrCLi2smPtguXsogV7vYjjLisSvcYXeYWOrk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=bY7/F9de; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="bY7/F9de" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff78bd3026so10585754a91.1 for ; Tue, 25 Mar 2025 17:09:38 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947778; x=1743552578; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=kF38Hh4HUGFV2sUBf6Ek+tmHe3WJPjIAvKIShcoR8VY=; b=bY7/F9deaBggQpWTaS7LnkaoKWVeenG0+i+QpGsjCVPbluAo4SVWK7EMVfp2FRWfWC F2ojNbEUbkMWqqzLQ4HArwaKn3QgP/IO1NX5X99m+2Lt5PWv0T9AyI5Ua8/n2dI1sY/8 p8+UiqbPeJR6liN09Fo+CdDWNnNTnvv2tRKOZJsLQ4ViKnFZjGkJRP1+y2L5lx6AIlOc BDGba/oIEuHy0DRk56KTg9T70PQR9QxsAcTvTdvfLgstF5PPGkVGcy4jpfMtFE6WBmho VjaRCMDFEPcBWLVGzsnAexYAVl2XN/aiQmz/nCdPXeT/HmWzk39BmmeIfPeFyb5XNnPn 9pRQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947778; x=1743552578; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=kF38Hh4HUGFV2sUBf6Ek+tmHe3WJPjIAvKIShcoR8VY=; b=IuU2en13KGXFXvBRlXUTbwJLr+yOZV2EmYYIkIuNiFIhJJpZDyuJ+7yPmC7JfnUCyC 89C19rhYV0A6yGqKVaLkU1QjMuFTuiohX9+KK4GEyH+GeCO/h9/WUxfSFfRmwG1Wr1Gx F34LGbA6tkj0YIIKudvF10GrFaUy9zt4KkCG6G/9ajaabaBHQxbHH2zRVzShUW8j0RVK GN+Bs45q4bx/B/n1tKbloqkiBGNbKEZdlNkf56HpLeZsd3WbNohfqf9+UB+baBeOQu3D PdPcsz1BDWzbpMkG0SuxlChLGif+dZwBSL72gqZQnDe+3MQquxZqHZndymPFpoDTBVrf APCw== X-Forwarded-Encrypted: i=1; AJvYcCV2mveywCRgpThde33myfLaE/1vPzasWtF7urOgsu8d3jdlG3Kb/3c2gEPSrItnpCnbQ7nQResT2wdTcQk=@vger.kernel.org X-Gm-Message-State: AOJu0YwyYgoRXKGEfORqJCVtA4DcYFhcJovx34S6NHLaRe3Pxvvl0Tbg sWQ3vgdBisGIPdGLzl0xa4E+ctlq/0KRJ21rCV3sYnJ8EVAvsRSoNlWeRNK562jkeKUpZw60M/T 3MzNrHA== X-Google-Smtp-Source: AGHT+IENV8oFSD151C97ymoORJVQv9ltJksDQxVX2GrApLN2FZNqFWOP19CLDalPGsJnpbadh2id0aHLjiDt X-Received: from pjur7.prod.google.com ([2002:a17:90a:d407:b0:2fa:1771:e276]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:2887:b0:2ee:edae:780 with SMTP id 98e67ed59e1d1-3030fea933fmr33662724a91.15.1742947778500; Tue, 25 Mar 2025 17:09:38 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:28 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-31-irogers@google.com> Subject: [PATCH v3 30/35] perf vendor events: Update skylakex events/metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topics, metrics to be generated from the TMA spreadsheet and other small clean ups. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/skylakex/cache.json | 74 ++++ .../pmu-events/arch/x86/skylakex/other.json | 74 ---- .../arch/x86/skylakex/skx-metrics.json | 385 +++++++++--------- 3 files changed, 266 insertions(+), 267 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/skylakex/cache.json b/tools/per= f/pmu-events/arch/x86/skylakex/cache.json index 2ce070629c52..7aeeb5725630 100644 --- a/tools/perf/pmu-events/arch/x86/skylakex/cache.json +++ b/tools/perf/pmu-events/arch/x86/skylakex/cache.json @@ -1,4 +1,78 @@ [ + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", + "SampleAfterValue": "2000003", + "UMask": "0x20" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDM", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDM", + "SampleAfterValue": "2000003", + "UMask": "0x10" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", + "SampleAfterValue": "2000003", + "UMask": "0x2" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITI", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITI", + "SampleAfterValue": "2000003", + "UMask": "0x1" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", + "SampleAfterValue": "2000003", + "UMask": "0x40" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDM", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDM", + "SampleAfterValue": "2000003", + "UMask": "0x8" + }, + { + "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", + "Counter": "0,1,2,3", + "EventCode": "0xEF", + "EventName": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", + "SampleAfterValue": "2000003", + "UMask": "0x4" + }, + { + "BriefDescription": "Counts number of cache lines that are dropped= and not written back to L3 as they are deemed to be less likely to be reus= ed shortly", + "Counter": "0,1,2,3", + "EventCode": "0xFE", + "EventName": "IDI_MISC.WB_DOWNGRADE", + "PublicDescription": "Counts number of cache lines that are droppe= d and not written back to L3 as they are deemed to be less likely to be reu= sed shortly.", + "SampleAfterValue": "100003", + "UMask": "0x4" + }, + { + "BriefDescription": "Counts number of cache lines that are allocat= ed and written back to L3 with the intention that they are more likely to b= e reused shortly", + "Counter": "0,1,2,3", + "EventCode": "0xFE", + "EventName": "IDI_MISC.WB_UPGRADE", + "PublicDescription": "Counts number of cache lines that are alloca= ted and written back to L3 with the intention that they are more likely to = be reused shortly.", + "SampleAfterValue": "100003", + "UMask": "0x2" + }, { "BriefDescription": "L1D data line replacements", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/skylakex/other.json b/tools/per= f/pmu-events/arch/x86/skylakex/other.json index 44c820518e12..adf7b6bb5838 100644 --- a/tools/perf/pmu-events/arch/x86/skylakex/other.json +++ b/tools/perf/pmu-events/arch/x86/skylakex/other.json @@ -35,62 +35,6 @@ "SampleAfterValue": "200003", "UMask": "0x40" }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDFE", - "SampleAfterValue": "2000003", - "UMask": "0x20" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IFWDM", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IFWDM", - "SampleAfterValue": "2000003", - "UMask": "0x10" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITFSE", - "SampleAfterValue": "2000003", - "UMask": "0x2" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_IHITI", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_IHITI", - "SampleAfterValue": "2000003", - "UMask": "0x1" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDFE", - "SampleAfterValue": "2000003", - "UMask": "0x40" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SFWDM", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SFWDM", - "SampleAfterValue": "2000003", - "UMask": "0x8" - }, - { - "BriefDescription": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", - "Counter": "0,1,2,3", - "EventCode": "0xEF", - "EventName": "CORE_SNOOP_RESPONSE.RSP_SHITFSE", - "SampleAfterValue": "2000003", - "UMask": "0x4" - }, { "BriefDescription": "Number of hardware interrupts received by the= processor.", "Counter": "0,1,2,3", @@ -100,24 +44,6 @@ "SampleAfterValue": "203", "UMask": "0x1" }, - { - "BriefDescription": "Counts number of cache lines that are dropped= and not written back to L3 as they are deemed to be less likely to be reus= ed shortly", - "Counter": "0,1,2,3", - "EventCode": "0xFE", - "EventName": "IDI_MISC.WB_DOWNGRADE", - "PublicDescription": "Counts number of cache lines that are droppe= d and not written back to L3 as they are deemed to be less likely to be reu= sed shortly.", - "SampleAfterValue": "100003", - "UMask": "0x4" - }, - { - "BriefDescription": "Counts number of cache lines that are allocat= ed and written back to L3 with the intention that they are more likely to b= e reused shortly", - "Counter": "0,1,2,3", - "EventCode": "0xFE", - "EventName": "IDI_MISC.WB_UPGRADE", - "PublicDescription": "Counts number of cache lines that are alloca= ted and written back to L3 with the intention that they are more likely to = be reused shortly.", - "SampleAfterValue": "100003", - "UMask": "0x2" - }, { "BriefDescription": "MEMORY_DISAMBIGUATION.HISTORY_RESET", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/skylakex/skx-metrics.json b/too= ls/perf/pmu-events/arch/x86/skylakex/skx-metrics.json index 2fe630cd4927..7cc7b076c3e2 100644 --- a/tools/perf/pmu-events/arch/x86/skylakex/skx-metrics.json +++ b/tools/perf/pmu-events/arch/x86/skylakex/skx-metrics.json @@ -295,12 +295,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED_PORT.PORT_0 + UOPS_DISPATCHED_PORT= .PORT_1 + UOPS_DISPATCHED_PORT.PORT_5 + UOPS_DISPATCHED_PORT.PORT_6) / tma_= info_thread_slots", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -312,7 +312,7 @@ "MetricExpr": "34 * (FP_ASSIST.ANY + OTHER_ASSISTS.ANY) / tma_info= _thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.AN= Y", "ScaleUnit": "100%" }, @@ -323,7 +323,7 @@ "MetricName": "tma_backend_bound", "MetricThreshold": "tma_backend_bound > 0.2", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound", + "PublicDescription": "This category represents fraction of slots w= here no uops are being delivered due to a lack of required resources for ac= cepting new uops in the Backend. Backend is the portion of the processor co= re where the out-of-order scheduler dispatches ready uops into their respec= tive execution units; and once completed these uops get retired according t= o program order. For example; stalls due to data-cache misses or stalls due= to the divider unit being overloaded are both categorized under Backend Bo= und. Backend Bound is further divided into two main categories: Memory Boun= d and Core Bound.", "ScaleUnit": "100%" }, { @@ -333,12 +333,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -353,7 +353,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -361,7 +361,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_dtlb_store)) + tma_memory_bound * (tma_store_bound / (tma_l1_bound= + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_= store_latency / (tma_store_latency + tma_false_sharing + tma_split_stores += tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency)) + tma_memory_bound * (tma_store_bound / (tma_dram_bou= nd + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_= store_latency / (tma_dtlb_store + tma_false_sharing + tma_split_stores + tm= a_store_latency)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -369,22 +369,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_b= ranches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tm= a_ms_switches + tma_lcp + tma_dsb_switches)) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * tma_fetch_latency * (tma_ms_swit= ches + tma_branch_resteers * (tma_clears_resteers + tma_mispredicts_resteer= s * (10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_misp= redicts)) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_b= ranches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + t= ma_itlb_misses + tma_lcp + tma_ms_switches)) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_mispredicts_resteers = + tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_i= tlb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_clea= rs_resteers + tma_mispredicts_resteers * (10 * tma_microcode_sequencer * tm= a_other_mispredicts / tma_branch_mispredicts)) / (tma_clears_resteers + tma= _mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma= _dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_swit= ches) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_m= ispredicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes = / tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_= bound * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_= 0) / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tm= a_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequence= r) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -392,7 +392,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -400,7 +400,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_mem + tma_remote_cache) + tma_l3_bound / (t= ma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bo= und) * tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_spl= it_stores + tma_dtlb_store - tma_store_latency)) + tma_machine_clears * (1 = - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) * tma_remote_cach= e / (tma_local_mem + tma_remote_cache + tma_remote_mem) + tma_l3_bound / (t= ma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_boun= d) * (tma_contested_accesses + tma_data_sharing) / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / = (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bo= und) * tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_= stores + tma_store_latency - tma_store_latency)) + tma_machine_clears * (1 = - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", @@ -408,7 +408,7 @@ }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -420,10 +420,10 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", @@ -445,8 +445,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -454,8 +454,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -463,7 +463,7 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, @@ -472,7 +472,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -480,33 +480,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((47.5 * tma_info_system_core_frequency - 3.5 * tma= _info_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (OFFCOR= E_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / (OFFCORE_RESPONSE.DEMAND= _DATA_RD.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.SN= OOP_HIT_WITH_FWD))) + (47.5 * tma_info_system_core_frequency - 3.5 * tma_in= fo_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_L= OAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(44 * tma_info_system_core_frequency * (MEM_LOAD_L3= _HIT_RETIRED.XSNP_HITM * (OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER= _CORE / (OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OFFCORE_R= ESPONSE.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) + 44 * tma_info_system_= core_frequency * MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS) * (1 + MEM_LOAD_RETIRED= .FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related= metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_fals= e_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_HITM_PS;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS_PS. Re= lated metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma= _false_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -517,25 +517,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(47.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_HIT + MEM_LOAD_= L3_HIT_RETIRED.XSNP_HITM * (1 - OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM= _OTHER_CORE / (OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE + OFF= CORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_R= ETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "44 * tma_info_system_core_frequency * (MEM_LOAD_L3_= HIT_RETIRED.XSNP_HIT + MEM_LOAD_L3_HIT_RETIRED.XSNP_HITM * (1 - OFFCORE_RES= PONSE.DEMAND_DATA_RD.L3_HIT.HITM_OTHER_CORE / (OFFCORE_RESPONSE.DEMAND_DATA= _RD.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.DEMAND_DATA_RD.L3_HIT.SNOOP_H= IT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / = 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT. Related metrics: tma_bottleneck_memory_synchron= ization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, tma= _remote_cache", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_HIT_PS. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -544,7 +544,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -554,7 +554,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -564,7 +564,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -572,27 +572,27 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(9 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(9 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -601,18 +601,18 @@ "MetricExpr": "(110 * tma_info_system_core_frequency * (OFFCORE_RE= SPONSE.DEMAND_RFO.L3_MISS.REMOTE_HITM + OFFCORE_RESPONSE.PF_L2_RFO.L3_MISS.= REMOTE_HITM) + 47.5 * tma_info_system_core_frequency * (OFFCORE_RESPONSE.DE= MAND_RFO.L3_HIT.HITM_OTHER_CORE + OFFCORE_RESPONSE.PF_L2_RFO.L3_HIT.HITM_OT= HER_CORE)) / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM, OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.HITM_OTHER_CORE. Rela= ted metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses,= tma_data_sharing, tma_machine_clears, tma_remote_cache", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: MEM_LOAD_L3_HIT_= RETIRED.XSNP_HITM_PS;OFFCORE_RESPONSE.DEMAND_RFO.L3_HIT.SNOOP_HITM. Related= metrics: tma_bottleneck_memory_synchronization, tma_contested_accesses, tm= a_data_sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric does a *rough estimation* of how = often L1D Fill Buffer unavailability limited additional L1D miss memory acc= ess requests to proceed", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D0x1@ / tma_info_thread_clks", + "MetricExpr": "tma_info_memory_load_miss_real_latency * cpu@L1D_PE= ND_MISS.FB_FULL\\,cmask\\=3D1@ / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;TopdownL4;tma_L4_group;tma_issueBW;t= ma_issueSL;tma_issueSmSt;tma_l1_bound_group", "MetricName": "tma_fb_full", "MetricThreshold": "tma_fb_full > 0.3", - "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy", + "PublicDescription": "This metric does a *rough estimation* of how= often L1D Fill Buffer unavailability limited additional L1D miss memory ac= cess requests to proceed. The higher the metric value; the deeper the memor= y hierarchy level the misses are satisfied from (metric values >1 are valid= ). Often it hints on approaching bandwidth limits (to L2 cache; L3 cache or= external memory). Related metrics: tma_bottleneck_cache_memory_bandwidth, = tma_info_system_dram_bw_use, tma_mem_bandwidth, tma_sq_full, tma_store_late= ncy, tma_streaming_stores", "ScaleUnit": "100%" }, { @@ -622,7 +622,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -632,7 +632,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -652,7 +652,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -661,7 +661,7 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { @@ -669,17 +669,17 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / UOPS_RETIRED.RETIRE_= SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { "BriefDescription": "This metric approximates arithmetic floating-= point (FP) vector uops fraction the CPU has retired aggregated across all v= ector widths", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umas= k\\=3D0xFC@ / UOPS_RETIRED.RETIRE_SLOTS", + "MetricExpr": "cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umas= k\\=3D0xfc@ / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -688,7 +688,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -697,7 +697,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -706,7 +706,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / UOPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -717,35 +717,35 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions , where one uop can represent mul= tiple contiguous instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring fused instructions -- where one uop can represent mu= ltiple contiguous instructions", "MetricExpr": "tma_light_operations * UOPS_RETIRED.MACRO_FUSED / U= OPS_RETIRED.RETIRE_SLOTS", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_fused_instructions", "MetricThreshold": "tma_fused_instructions > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions , where one uop can represent mu= ltiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of l= egacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Oth= er_Light_Ops in MTL!)}", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring fused instructions -- where one uop can represent m= ultiple contiguous instructions. CMP+JCC or DEC+JCC are common examples of = legacy fusions. {([MTL] Note new MOV+OP and Load+OP fusions appear under Ot= her_Light_Ops in MTL!)}", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", "MetricExpr": "(UOPS_RETIRED.RETIRE_SLOTS + UOPS_RETIRED.MACRO_FUS= ED - INST_RETIRED.ANY) / tma_info_thread_slots", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to instruction cache misses", - "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D0x1\\,edge\\=3D0x1@) / tma_info_thread_clks", + "MetricExpr": "(ICACHE_16B.IFDATA_STALL + 2 * cpu@ICACHE_16B.IFDAT= A_STALL\\,cmask\\=3D1\\,edge@) / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -756,11 +756,11 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "tma_info_inst_mix_instructions / (UOPS_RETIRED.RETI= RE_SLOTS / UOPS_ISSUED.ANY * BR_MISP_EXEC.INDIRECT)", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { "BriefDescription": "Number of Instructions per non-speculative Br= anch Misprediction (JEClear) (lower number means higher occurrence rate)", @@ -785,7 +785,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= )))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_mite= )))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -794,7 +794,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_mite))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -802,10 +802,11 @@ }, { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -834,7 +835,7 @@ }, { "BriefDescription": "Core actual clocks when any Logical Processor= is active on the Physical Core", - "MetricExpr": "(CPU_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tm= a_info_thread_clks)", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / 2 * (1 + CPU_CLK_UNHALTE= D.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_XCLK) if #core_wide < 1 else (CP= U_CLK_UNHALTED.THREAD_ANY / 2 if #SMT_on else tma_info_thread_clks))", "MetricGroup": "SMT", "MetricName": "tma_info_core_core_clks" }, @@ -859,14 +860,14 @@ }, { "BriefDescription": "Actual per-core usage of the Floating Point n= on-X87 execution units (regardless of precision or vector-width)", - "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + cpu@FP_ARITH_INST_R= ETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xFC@) / (2 * tma_info_core_core_clks= )", + "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + cpu@FP_ARITH_INST_R= ETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xfc@) / (2 * tma_info_core_core_clks= )", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -879,20 +880,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / DSB2MITE_SWITCHE= S.COUNT", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@ + 2", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@ + 2", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -928,7 +929,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -943,11 +944,11 @@ { "BriefDescription": "Instructions per FP Arithmetic instruction (l= ower number means higher occurrence rate)", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xFC@)", + "MetricExpr": "INST_RETIRED.ANY / (FP_ARITH_INST_RETIRED.SCALAR + = cpu@FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE\\,umask\\=3D0xfc@)", "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -955,7 +956,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -963,7 +964,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -971,7 +972,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -979,7 +980,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -987,7 +988,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -1037,7 +1038,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 4 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 9", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1224,8 +1225,8 @@ "MetricName": "tma_info_memory_tlb_store_stlb_mpki" }, { - "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per core", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "BriefDescription": "", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1246,12 +1247,12 @@ "MetricExpr": "INST_RETIRED.ANY / (FP_ASSIST.ANY + OTHER_ASSISTS.A= NY)", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / cpu@UOPS_RETIRED.RETIRE= _SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1307,14 +1308,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1332,7 +1332,7 @@ }, { "BriefDescription": "Average number of parallel data read requests= to external memory", - "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@UNC_CHA_TOR= _OCCUPANCY.IA_MISS_DRD\\,thresh\\=3D0x1@", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_OCC= UPANCY.IA_MISS_DRD@thresh\\=3D1@", "MetricGroup": "Mem;MemoryBW;SoC", "MetricName": "tma_info_system_mem_parallel_reads", "PublicDescription": "Average number of parallel data read request= s to external memory. Accounts for demand loads and L1/L2 prefetches" @@ -1362,7 +1362,7 @@ "MetricExpr": "(CORE_POWER.LVL0_TURBO_LICENSE / 2 / tma_info_core_= core_clks if #SMT_on else CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_cor= e_clks)", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1370,7 +1370,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1378,7 +1378,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1412,7 +1412,7 @@ "MetricName": "tma_info_system_uncore_frequency" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1421,15 +1421,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1455,15 +1454,15 @@ "MetricExpr": "UOPS_RETIRED.RETIRE_SLOTS / BR_INST_RETIRED.NEAR_TA= KEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 4 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 6" }, { "BriefDescription": "This metric represents fraction of cycles the= CPU was stalled due to Instruction TLB (ITLB) misses", "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1471,7 +1470,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1480,17 +1479,17 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates how often the CPU was s= talled due to L2 cache accesses by loads", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D0x1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2= _MISS) / tma_info_thread_clks)", + "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + cpu@L1D_PEND_MISS.FB_FULL\\,cm= ask\\=3D1@) * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_M= ISS) / tma_info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1499,7 +1498,7 @@ "MetricExpr": "3.5 * tma_info_system_core_frequency * MEM_LOAD_RET= IRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) = / tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1508,17 +1507,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(20.5 * tma_info_system_core_frequency - 3.5 * tma_= info_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETI= RED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "17 * tma_info_system_core_frequency * (MEM_LOAD_RET= IRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2))= / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1526,18 +1525,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "tma_retiring - tma_heavy_operations", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1555,7 +1554,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1563,39 +1562,39 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from local memory", - "MetricExpr": "(80 * tma_info_system_core_frequency - 20.5 * tma_i= nfo_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.LOCAL_DRAM * (1 + MEM= _LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks= ", + "MetricExpr": "59.5 * tma_info_system_core_frequency * MEM_LOAD_L3= _MISS_RETIRED.LOCAL_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;TopdownL5;tma_L5_group;tma_mem_latency_grou= p", "MetricName": "tma_local_mem", - "MetricThreshold": "tma_local_mem > 0.1 & tma_mem_latency > 0.1 & = tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_local_mem > 0.1 & (tma_mem_latency > 0.1 &= (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2)= ))", "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from local memory. Caching will = improve the latency and increase performance. Sample with: MEM_LOAD_L3_MISS= _RETIRED.LOCAL_DRAM", "ScaleUnit": "100%" }, @@ -1604,7 +1603,7 @@ "MetricExpr": "(12 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (11= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1621,10 +1620,10 @@ }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1633,7 +1632,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1645,11 +1644,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", "MetricName": "tma_memory_operations", @@ -1671,7 +1670,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1685,12 +1684,12 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { @@ -1698,7 +1697,7 @@ "MetricExpr": "2 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1708,7 +1707,7 @@ "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_non_fused_branches", "MetricThreshold": "tma_non_fused_branches > 0.1 & tma_light_opera= tions > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring branch instructions that were not fused. Non-condit= ional branches like direct JMP or CALL would count here. Can be used to exa= mine fusible conditional jumps that were not fused.", "ScaleUnit": "100%" }, { @@ -1716,8 +1715,8 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / UOPS_RETI= RED.RETIRE_SLOTS", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP_PS", "ScaleUnit": "100%" }, { @@ -1730,19 +1729,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1751,7 +1750,7 @@ "MetricGroup": "Compute;TopdownL6;tma_L6_group;tma_alu_op_utilizat= ion_group;tma_issue2P", "MetricName": "tma_port_0", "MetricThreshold": "tma_port_0 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED_PORT.PORT_0. Related metrics: tma_fp_= scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vecto= r_512b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 0 ([SNB+] ALU; [HSW+] ALU and 2nd = branch). Sample with: UOPS_DISPATCHED.PORT_0. Related metrics: tma_fp_scala= r, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512= b, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1760,7 +1759,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_1", "MetricThreshold": "tma_port_1 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED_PORT.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vect= or_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tm= a_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 1 (ALU). Sample with: UOPS_DISPATC= HED.PORT_1. Related metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_12= 8b, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_5, tma_por= t_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1796,7 +1795,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_5", "MetricThreshold": "tma_port_5 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_5. Related metrics: tma_fp_sc= alar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_= 512b, tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 5 ([SNB+] Branches and ALU; [HSW+]= ALU). Sample with: UOPS_DISPATCHED.PORT_5. Related metrics: tma_fp_scalar,= tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b,= tma_port_0, tma_port_1, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1805,7 +1804,7 @@ "MetricGroup": "TopdownL6;tma_L6_group;tma_alu_op_utilization_grou= p;tma_issue2P", "MetricName": "tma_port_6", "MetricThreshold": "tma_port_6 > 0.6", - "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED_PORT.PORT_1. Related metrics: tma_fp_s= calar, tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector= _512b, tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", + "PublicDescription": "This metric represents Core fraction of cycl= es CPU dispatched uops on execution port 6 ([HSW+] Primary Branch and simpl= e ALU). Sample with: UOPS_DISPATCHED.PORT_1. Related metrics: tma_fp_scalar= , tma_fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b= , tma_port_0, tma_port_1, tma_port_5, tma_ports_utilized_2", "ScaleUnit": "100%" }, { @@ -1822,8 +1821,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1831,8 +1830,8 @@ "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_cl= ks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1840,7 +1839,7 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_1 - UOPS_EXECUTED.CO= RE_CYCLES_GE_2) / 2 if #SMT_on else EXE_ACTIVITY.1_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Related metrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1849,35 +1848,35 @@ "MetricExpr": "((UOPS_EXECUTED.CORE_CYCLES_GE_2 - UOPS_EXECUTED.CO= RE_CYCLES_GE_3) / 2 if #SMT_on else EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_c= ore_core_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. R= elated metrics: tma_fp_scalar, tma_fp_vector, tma_fp_vector_128b, tma_fp_ve= ctor_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, tma_port= _6", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise)", + "BriefDescription": "This metric represents fraction of cycles CPU= executed total of 3 or more uops per cycle on all execution ports (Logical= Processor cycles since ICL, Physical Core cycles otherwise).", "MetricExpr": "(UOPS_EXECUTED.CORE_CYCLES_GE_3 / 2 if #SMT_on else= UOPS_EXECUTED.CORE_CYCLES_GE_3) / tma_info_core_core_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote cache in other socket= s including synchronizations issues", "MetricConstraint": "NO_GROUP_EVENTS_NMI", - "MetricExpr": "((110 * tma_info_system_core_frequency - 20.5 * tma= _info_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM + (110 = * tma_info_system_core_frequency - 20.5 * tma_info_system_core_frequency) *= MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_= LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", + "MetricExpr": "(89.5 * tma_info_system_core_frequency * MEM_LOAD_L= 3_MISS_RETIRED.REMOTE_HITM + 89.5 * tma_info_system_core_frequency * MEM_LO= AD_L3_MISS_RETIRED.REMOTE_FWD) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RE= TIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Offcore;Server;Snoop;TopdownL5;tma_L5_group;tma_is= sueSyncxn;tma_mem_latency_group", "MetricName": "tma_remote_cache", - "MetricThreshold": "tma_remote_cache > 0.05 & tma_mem_latency > 0.= 1 & tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2= ", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. Sample with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_HITM, MEM= _LOAD_L3_MISS_RETIRED.REMOTE_FWD. Related metrics: tma_bottleneck_memory_sy= nchronization, tma_contested_accesses, tma_data_sharing, tma_false_sharing,= tma_machine_clears", + "MetricThreshold": "tma_remote_cache > 0.05 & (tma_mem_latency > 0= .1 & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > = 0.2)))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote cache in other socke= ts including synchronizations issues. This is caused often due to non-optim= al NUMA allocations. #link to NUMA article. Sample with: MEM_LOAD_L3_MISS_R= ETIRED.REMOTE_HITM_PS;MEM_LOAD_L3_MISS_RETIRED.REMOTE_FWD_PS. Related metri= cs: tma_bottleneck_memory_synchronization, tma_contested_accesses, tma_data= _sharing, tma_false_sharing, tma_machine_clears", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling loads from remote memory", - "MetricExpr": "(147.5 * tma_info_system_core_frequency - 20.5 * tm= a_info_system_core_frequency) * MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM * (1 += MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_= clks", + "MetricExpr": "127 * tma_info_system_core_frequency * MEM_LOAD_L3_= MISS_RETIRED.REMOTE_DRAM * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.= L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "Server;Snoop;TopdownL5;tma_L5_group;tma_mem_latenc= y_group", "MetricName": "tma_remote_mem", - "MetricThreshold": "tma_remote_mem > 0.1 & tma_mem_latency > 0.1 &= tma_dram_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. Sample with: MEM_LOAD_L3_MISS= _RETIRED.REMOTE_DRAM", + "MetricThreshold": "tma_remote_mem > 0.1 & (tma_mem_latency > 0.1 = & (tma_dram_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling loads from remote memory. This is caus= ed often due to non-optimal NUMA allocations. #link to NUMA article. Sample= with: MEM_LOAD_L3_MISS_RETIRED.REMOTE_DRAM_PS", "ScaleUnit": "100%" }, { @@ -1895,7 +1894,7 @@ "MetricExpr": "PARTIAL_RAT_STALLS.SCOREBOARD / tma_info_thread_clk= s", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: PARTIAL_RAT_STALLS.SCOREBOARD. Related me= trics: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1906,7 +1905,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1914,8 +1913,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES. Related metrics: tma_port_4", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1923,7 +1922,7 @@ "MetricExpr": "(OFFCORE_REQUESTS_BUFFER.SQ_FULL / 2 if #SMT_on els= e OFFCORE_REQUESTS_BUFFER.SQ_FULL) / tma_info_core_core_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1932,8 +1931,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1941,8 +1940,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1951,8 +1950,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 11 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1968,7 +1967,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -1976,31 +1975,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -2008,7 +2007,7 @@ "MetricExpr": "9 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -2017,8 +2016,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 1DA7818A6B5 for ; Wed, 26 Mar 2025 00:09:47 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947791; cv=none; b=h1vKL4UN3gPUYUl1SaKr+HYsICvnqBZa+DrcKSa9TIWcKxe2FQl3Wi2ukxq/c91Ug7/Yqpoi3QJ7UYSdLPbjXe5qjsWyCdObdxMtQkc2mlw14vPzUu2eLCVeqMQUmO6Nkd6Gb7PlbbUBLVhZWV5qtKMEp/oItXFaoONRQQ9b4q4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947791; c=relaxed/simple; bh=0AT9S4+tQKLFJnJKnyWkGKU0/cg5F7/LVzqrzS1ky+g=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=U2xEGYVDYw7T2+JpfyHdekvSC1KKZ1io3tmwLS6n4QvMAhiSTHqXVQXeTYGYBJedatSdlUN/T2eT9u3ZH2u8rN7wopnRFZPw+LSroqVcMJs2qWDaKUgDTdo4IOMbrO+5851GBfHmwBfLRNsKcMEeCJ6/vxCnS4Vo/3ctbxn4QiA= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=ctFKppnO; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="ctFKppnO" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2ff58318acaso17420068a91.0 for ; Tue, 25 Mar 2025 17:09:47 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947787; x=1743552587; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=umn9Nu09RuoBihtyilxpcZ9ZnsjZo8cvNgml2LAfPCY=; b=ctFKppnOfHrQ0fyjQMIeNpX4zESAOPG6WBGG8JoiH5cUy4SdLS3QBxrwYYQONpurGw 6WDFVOpiPYjqwbwLivTSSIJqRlh5tluo+XN48cZVVB2GAtBPanFaHj0b4b3ciNbXGNH8 nX+Gu8LHjbOrBTrFQ14623bwwKgCkRT6yFDst7pmBB4VUJae8HXuU4PzZc9bPXqnTMDJ k+j2VbqBGs+47YuSbriqg2oDXfEhMNGbQxi0tQGGkuEWYLJdoil4mmN5qAh0RKbt47xf rtULzDk5VIkBrXFMUkJBpIjp6gEjwLOuatanAFHEWM3iFSWjPhac63rBr1QO9Y4rdGdG /p5Q== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947787; x=1743552587; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=umn9Nu09RuoBihtyilxpcZ9ZnsjZo8cvNgml2LAfPCY=; b=AtjncYC6fOx1ujzjLp8Z4wiHDPejSlip6+SddxTTR66ryJRSYdGUNgrlZCF3d99W/U XPfSSZ4EOtK9IGPSnSRvcatvBqxQmW47KZDpsY2f775ABOXOYVepD2Q768ccFIGv2OFO JB2/70GFI1FOtR88+L4WOKBzGQc3SHUiL8HG3PGz+p+/smjlGsQTQOug8V15fpgPHHSX /mV+cs0lPGu0Ujrp/5ukNzE3L0RuLkWVhaI9dKLQ7+RACp8Mh7le0Mh7kgi+Lk6CTMIk ARbXCDpZxxqydMe4G+/ziA+AsVhXrfmPEDT+qXCqGTemNpQjeXmfehqnFIDV6bgOalg3 tUDA== X-Forwarded-Encrypted: i=1; AJvYcCW24j4X4SBeTDRkpL706m7XRzzB+mvCVXZNmxU6TG3XTkdczMZNAbhAG4AjACKuG+3V0Ay6MC3WjVoM7aE=@vger.kernel.org X-Gm-Message-State: AOJu0Yz/Jkgfw7aUCt6iBtcD6YRXYT+2yKbs/g1M5d0f3SL6cIJfY46N qbg7wkW6gGQlXlj+Q5cuEmLMJGWwRMEZCV6uD82k4dMNP5Mu7xSePF+dnqVvlY90LaablQhF8Kt axyeGeA== X-Google-Smtp-Source: AGHT+IG7tJBdLQt+JiiRjeh94pcBEFZUxUWYxH+OLGhKxd2B4HZuBNfutj/wwUkHFqT45oiQ8ANAONPbAu+r X-Received: from pjbqd16.prod.google.com ([2002:a17:90b:3cd0:b0:2ef:8055:93d9]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:2647:b0:2ff:592d:23bc with SMTP id 98e67ed59e1d1-3030fe58cc6mr30094855a91.4.1742947787303; Tue, 25 Mar 2025 17:09:47 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:29 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-32-irogers@google.com> Subject: [PATCH v3 31/35] perf vendor events: Update snowridgex events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and memory. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/snowridgex/cache.json | 192 +++++++++ .../arch/x86/snowridgex/memory.json | 202 +++++++++ .../pmu-events/arch/x86/snowridgex/other.json | 394 ------------------ 3 files changed, 394 insertions(+), 394 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/snowridgex/cache.json b/tools/p= erf/pmu-events/arch/x86/snowridgex/cache.json index 7882dca9d5e1..1bb42acf1d48 100644 --- a/tools/perf/pmu-events/arch/x86/snowridgex/cache.json +++ b/tools/perf/pmu-events/arch/x86/snowridgex/cache.json @@ -357,6 +357,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.COREWB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x3000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -367,6 +377,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have an outstanding request. Returns the number of cycles unt= il the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.COREWB_M.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8003000000000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -427,6 +457,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -487,6 +527,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve an outstanding request. Returns the number of cycles until the response = is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.L3_HIT", "Counter": "0,1,2,3", @@ -553,6 +614,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y the L3 cache.", "Counter": "0,1,2,3", @@ -613,6 +695,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have an outstan= ding request. Returns the number of cycles until the response is received (= i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -623,6 +715,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that have any type of resp= onse.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10400", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that were supplied by the = L3 cache where a snoop was sent, the snoop hit, and modified data was forwa= rded.", "Counter": "0,1,2,3", @@ -633,6 +735,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -693,6 +805,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have an outstanding request. Returns th= e number of cycles until the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -753,6 +885,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -813,6 +955,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have an outstanding request. Returns the numb= er of cycles until the response is received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.L1WB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x1000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -823,6 +985,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that have any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.L2WB_M.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x2000000010000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that were supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -843,6 +1015,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e any type of response.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x10477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -903,6 +1085,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e an outstanding request. Returns the number of cycles until the response i= s received (i.e. XQ to XQ latency).", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.OUTSTANDING", + "MSRIndex": "0x1a6", + "MSRValue": "0x8000000000000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that were supplied by= the L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/snowridgex/memory.json b/tools/= perf/pmu-events/arch/x86/snowridgex/memory.json index 34306ec24e9b..260a488540bb 100644 --- a/tools/perf/pmu-events/arch/x86/snowridgex/memory.json +++ b/tools/perf/pmu-events/arch/x86/snowridgex/memory.json @@ -25,6 +25,16 @@ "SampleAfterValue": "200003", "UMask": "0x4" }, + { + "BriefDescription": "Counts all code reads that were supplied by D= RAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.ALL_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000044", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all code reads that were not supplied = by the L3 cache.", "Counter": "0,1,2,3", @@ -45,6 +55,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all code reads that were supplied by D= RAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.ALL_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000044", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -65,6 +85,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -85,6 +115,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000004", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -105,6 +155,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.L3_MISS", "Counter": "0,1,2,3", @@ -127,6 +198,27 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", + "Counter": "0,1,2,3", + "Deprecated": "1", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000001", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were not suppli= ed by the L3 cache.", "Counter": "0,1,2,3", @@ -147,6 +239,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000002", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -167,6 +269,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -187,6 +299,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_CODE_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000040", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -207,6 +339,26 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000010", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -227,6 +379,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000020", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that were not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -317,6 +479,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e not supplied by the L3 cache.", "Counter": "0,1,2,3", @@ -337,6 +509,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x184000477", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts streaming stores that were not supplie= d by the L3 cache.", "Counter": "0,1,2,3", @@ -357,6 +539,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.UC_RD.DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100184000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts uncached memory reads that were not su= pplied by the L3 cache.", "Counter": "0,1,2,3", @@ -377,6 +569,16 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, + { + "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", + "Counter": "0,1,2,3", + "EventCode": "0XB7", + "EventName": "OCR.UC_RD.LOCAL_DRAM", + "MSRIndex": "0x1a6,0x1a7", + "MSRValue": "0x100184000000", + "SampleAfterValue": "100003", + "UMask": "0x1" + }, { "BriefDescription": "Counts uncached memory writes that were not s= upplied by the L3 cache.", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/snowridgex/other.json b/tools/p= erf/pmu-events/arch/x86/snowridgex/other.json index 57613207f7ad..35cdbfa617e7 100644 --- a/tools/perf/pmu-events/arch/x86/snowridgex/other.json +++ b/tools/perf/pmu-events/arch/x86/snowridgex/other.json @@ -116,26 +116,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all code reads that were supplied by D= RAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.ALL_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000044", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all code reads that were supplied by D= RAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.ALL_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000044", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts all code reads that have an outstandin= g request. Returns the number of cycles until the response is received (i.e= . XQ to XQ latency).", "Counter": "0,1,2,3", @@ -146,180 +126,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.COREWB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x3000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writebacks from L1 cache and = L2 cache that have an outstanding request. Returns the number of cycles unt= il the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.COREWB_M.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8003000000000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand instruction fetches and L1 inst= ruction cache prefetches that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000004", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that we= re supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts cacheable demand data reads, L1 data c= ache hardware prefetches and software prefetches (except PREFETCHW) that ha= ve an outstanding request. Returns the number of cycles until the response = is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.ANY_RESPONSE", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.DRAM", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.LOCAL_DRAM", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "This event is deprecated. Refer to new event = OCR.DEMAND_DATA_AND_L1PF_RD.OUTSTANDING", - "Counter": "0,1,2,3", - "Deprecated": "1", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_DATA_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000001", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have any type o= f response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that were supplied b= y DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts demand reads for ownership (RFO) and s= oftware prefetches for exclusive ownership (PREFETCHW) that have an outstan= ding request. Returns the number of cycles until the response is received (= i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.DEMAND_RFO.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000002", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores which modify a full 6= 4 byte cacheline that have any type of response.", "Counter": "0,1,2,3", @@ -330,146 +136,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts L1 data cache hardware prefetches and = software prefetches (except PREFETCHW and PFRFO) that have any type of resp= onse.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L1D_AND_SWPF.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10400", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch code reads = (written to the L2 cache only) that have an outstanding request. Returns th= e number of cycles until the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_CODE_RD.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000040", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch data reads = (written to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_DATA_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000010", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that were supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts L2 cache hardware prefetch RFOs (writt= en to the L2 cache only) that have an outstanding request. Returns the numb= er of cycles until the response is received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.HWPF_L2_RFO.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000020", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writebacks from L1 cache that= miss the L2 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.L1WB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x1000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts modified writeBacks from L2 cache that= miss the L3 cache that have any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.L2WB_M.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x2000000010000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts miscellaneous requests, such as I/O ac= cesses, that have any type of response.", "Counter": "0,1,2,3", @@ -500,46 +166,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e any type of response.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.ANY_RESPONSE", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x10477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that wer= e supplied by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x184000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts all data read, code read and RFO reque= sts including demands and prefetches to the core caches (L1 or L2) that hav= e an outstanding request. Returns the number of cycles until the response i= s received (i.e. XQ to XQ latency).", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.READS_TO_CORE.OUTSTANDING", - "MSRIndex": "0x1a6", - "MSRValue": "0x8000000000000477", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts streaming stores that have any type of= response.", "Counter": "0,1,2,3", @@ -560,26 +186,6 @@ "SampleAfterValue": "100003", "UMask": "0x1" }, - { - "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.UC_RD.DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100184000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, - { - "BriefDescription": "Counts uncached memory reads that were suppli= ed by DRAM.", - "Counter": "0,1,2,3", - "EventCode": "0XB7", - "EventName": "OCR.UC_RD.LOCAL_DRAM", - "MSRIndex": "0x1a6,0x1a7", - "MSRValue": "0x100184000000", - "SampleAfterValue": "100003", - "UMask": "0x1" - }, { "BriefDescription": "Counts uncached memory reads that have an out= standing request. Returns the number of cycles until the response is receiv= ed (i.e. XQ to XQ latency).", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 7C9F6182BC for ; Wed, 26 Mar 2025 00:09:55 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947803; cv=none; b=YWAwGqZJzqsvSbBy1AB3N4TAx3FFYkch8MGd76t9v+WL7daxk3RZbN1rclMOfk2Pe6fZPgVffrTl/fwxnMKW08kB+nPftFx3ibWWw7yv025IBXwcUmbGocFpTXOSqg936/F1UJ/uJljJ7y6qgsIWYKa+gr6ILJLMK745Q2o4HEg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947803; c=relaxed/simple; bh=NI07KYNDZ+eoog7me4jDr2tImkQxoQ/XfiHMbFV8adg=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=MlF+IUqXj63xIhP86WunrIr2Sne/wiGbHfGqCaHM5L7brIXBOM2+ua6NVS5ak7VkHRxxXcSnxltrJWVK8Jrfwnl23ka2Yhx11HH9vfjtpfH6f3t+UM6ngmKryCm0A0mi99gXXGsrEILwcsjB9FeusPT0Tn6CwS5Ck5sRhPlwiSo= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=TdZ7fYAG; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="TdZ7fYAG" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-3032f4ea8cfso5895721a91.3 for ; Tue, 25 Mar 2025 17:09:55 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947795; x=1743552595; darn=vger.kernel.org; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:from:to:cc:subject:date:message-id :reply-to; bh=ycfq65OQXWk86ftKFh9rcavRUbjHb+EOA1eMhpCwt3E=; b=TdZ7fYAGaypkNPffXjsqoTJaDXezPOyMf2MFnr+ygTx08bI81BXtbhJEr4owvDTUyE cdCSza/HFWfptZn+jqqcOyOz1HJCiP7/cAf41jDtNtwiFOhtI4W+x682+unSBAcGDrCZ Vlg6t6z+oVJbz0FhaAQiD/gyoHUwuHV1PJR4Kwwl17Uon8A2inMKsVKnW9PCMtosoiwB qJjnPCAeg29/3+BkuEy4oDhqS4kMEiQ1PSNSZR77MKdLyOrWYT6jYf6zVh34//v5oKRV qs830UesivaGTb45I8ZTuaasPMAhFrGeHL9USI7Wc+ML/lGc9EsQyL1kg9CkxXy4cwRU DWLg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947795; x=1743552595; h=content-transfer-encoding:to:from:subject:message-id:references :mime-version:in-reply-to:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to; bh=ycfq65OQXWk86ftKFh9rcavRUbjHb+EOA1eMhpCwt3E=; b=M8sEX9YO89hdUM1iWak8PL5ZnXMM0dtQzTFfQDb6n/tP5GQDyOnVC3KYYX7GRonq2b cNKdbZK5jTyO6Ky6/DecI0obHVRPcs2q0nos8yGL+0nGm/8fufQJrGimBXkBmExiqVji cEbl3BXo3jh84hRGhFmNjRWRjNRO3KsAu+6QiHigkoGq7FcrCSVv62U4Bx1Gx0S8UPYr Gb+3W55BJFDZt5RVP+Jlu7ZNXhTyJy+R7x5xWRQ12kCO+ZzuN+Peo9pVkLLirShrfIJv z8WMrbw8YRp6EffgDvs9KiCnNC26WOQQzAOwqUIOptBHJR+uYtEtzJeHD/xiBNy2Ik1b 23Ew== X-Forwarded-Encrypted: i=1; AJvYcCVflNZlb8nxhAqD38cpbzdPElI+KGAPE6G5jZ80DOcwRd9tf3/eGXW3xulmKS6nZiYiZrNePhhElNZoq5k=@vger.kernel.org X-Gm-Message-State: AOJu0Yw79e/zehOYygOCCPcGCcPdXEIvO6mr1OoCFpyu3FL2t2KxkpkM epnIhNG4OpfcH0RRbzVMOdzJWfBFHzO4Aztkel4EB7Px+SVE2QjQeBNF7sEzPghGkb6egjLoHVO c0mO7Aw== X-Google-Smtp-Source: AGHT+IFq/BUlt4Qo2+pThjQ8++jRDO5+99gXU6XQXqkF09RKao8E7ICPQBduyMdpqlMzgLLZ4UU/k+AZi13r X-Received: from pgac11.prod.google.com ([2002:a05:6a02:294b:b0:af9:1196:c39c]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:502:b0:1f5:8714:8147 with SMTP id adf61e73a8af0-1fe42fd4962mr32396605637.23.1742947794974; Tue, 25 Mar 2025 17:09:54 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:30 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-33-irogers@google.com> Subject: [PATCH v3 32/35] perf vendor events: Update tigerlake metrics From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Switch to metrics generated from the TMA spreadsheet. Minor threshold simplification. Signed-off-by: Ian Rogers --- .../arch/x86/tigerlake/tgl-metrics.json | 383 +++++++++--------- 1 file changed, 191 insertions(+), 192 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/tigerlake/tgl-metrics.json b/to= ols/perf/pmu-events/arch/x86/tigerlake/tgl-metrics.json index 8c0cd6e63a2a..2db7a70f7a07 100644 --- a/tools/perf/pmu-events/arch/x86/tigerlake/tgl-metrics.json +++ b/tools/perf/pmu-events/arch/x86/tigerlake/tgl-metrics.json @@ -89,12 +89,12 @@ "MetricExpr": "LD_BLOCKS_PARTIAL.ADDRESS_ALIAS / tma_info_thread_c= lks", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_4k_aliasing", - "MetricThreshold": "tma_4k_aliasing > 0.2 & tma_l1_bound > 0.1 & t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound)", + "MetricThreshold": "tma_4k_aliasing > 0.2 & (tma_l1_bound > 0.1 & = (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates how often memory load = accesses were aliased by preceding stores (in program order) with a 4K addr= ess offset. False match is possible; which incur a few cycles load re-issue= . However; the short re-issue duration is often hidden by the out-of-order = core and HW optimizations; hence a user may safely ignore a high value of t= his metric unless it manages to propagate up into parent nodes of the hiera= rchy (e.g. to L1_Bound).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations", + "BriefDescription": "This metric represents Core fraction of cycle= s CPU dispatched uops on execution ports for ALU operations.", "MetricExpr": "(UOPS_DISPATCHED.PORT_0 + UOPS_DISPATCHED.PORT_1 + = UOPS_DISPATCHED.PORT_5 + UOPS_DISPATCHED.PORT_6) / (4 * tma_info_core_core_= clks)", "MetricGroup": "TopdownL5;tma_L5_group;tma_ports_utilized_3m_group= ", "MetricName": "tma_alu_op_utilization", @@ -106,7 +106,7 @@ "MetricExpr": "34 * ASSISTS.ANY / tma_info_thread_slots", "MetricGroup": "BvIO;TopdownL4;tma_L4_group;tma_microcode_sequence= r_group", "MetricName": "tma_assists", - "MetricThreshold": "tma_assists > 0.1 & tma_microcode_sequencer > = 0.05 & tma_heavy_operations > 0.1", + "MetricThreshold": "tma_assists > 0.1 & (tma_microcode_sequencer >= 0.05 & tma_heavy_operations > 0.1)", "PublicDescription": "This metric estimates fraction of slots the = CPU retired uops delivered by the Microcode_Sequencer as a result of Assist= s. Assists are long sequences of uops that are required in certain corner-c= ases for operations that cannot be handled natively by the execution pipeli= ne. For example; when working with very small floating point values (so-cal= led Denormals); the FP units are not set up to perform these operations nat= ively. Instead; a sequence of instructions to perform the computation on th= e Denormals is injected into the pipeline. Since these microcode sequences = might be dozens of uops long; Assists can be extremely deleterious to perfo= rmance and they can be avoided in many cases. Sample with: ASSISTS.ANY", "ScaleUnit": "100%" }, @@ -129,12 +129,12 @@ "MetricName": "tma_bad_speculation", "MetricThreshold": "tma_bad_speculation > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example", + "PublicDescription": "This category represents fraction of slots w= asted due to incorrect speculations. This include slots used to issue uops = that do not eventually get retired and slots for which the issue-pipeline w= as blocked due to recovery from earlier incorrect speculation. For example;= wasted work due to miss-predicted branches are categorized under Bad Specu= lation category. Incorrect data speculation followed by Memory Ordering Nuk= es is another example.", "ScaleUnit": "100%" }, { "BriefDescription": "Total pipeline cost of instruction fetch rela= ted bottlenecks by large code footprint programs (i-side cache; TLB and BTB= misses)", - "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_icache_misses + tma_itlb_misses = + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches)", + "MetricExpr": "100 * tma_fetch_latency * (tma_itlb_misses + tma_ic= ache_misses + tma_unknown_branches) / (tma_branch_resteers + tma_dsb_switch= es + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches)", "MetricGroup": "BigFootprint;BvBC;Fed;Frontend;IcMiss;MemoryTLB", "MetricName": "tma_bottleneck_big_code", "MetricThreshold": "tma_bottleneck_big_code > 20" @@ -149,7 +149,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Bandwidth related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dr= am_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound += tma_store_bound)) * (tma_fb_full / (tma_dtlb_load + tma_store_fwd_blk + tm= a_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4k_alias= ing + tma_fb_full)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_bandwidth / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_b= ound * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_= l3_bound + tma_store_bound)) * (tma_sq_full / (tma_contested_accesses + tma= _data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound * (tm= a_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound += tma_store_bound)) * (tma_fb_full / (tma_4k_aliasing + tma_dtlb_load + tma_= fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + = tma_store_fwd_blk)))", "MetricGroup": "BvMB;Mem;MemoryBW;Offcore;tma_issueBW", "MetricName": "tma_bottleneck_cache_memory_bandwidth", "MetricThreshold": "tma_bottleneck_cache_memory_bandwidth > 20", @@ -157,7 +157,7 @@ }, { "BriefDescription": "Total pipeline cost of external Memory- or Ca= che-Latency related bottlenecks", - "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_dtlb_load + tma_store_fwd_blk + tma_l1_latency_de= pendency + tma_lock_latency + tma_split_loads + tma_4k_aliasing + tma_fb_fu= ll)) + tma_memory_bound * (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tm= a_l3_bound + tma_dram_bound + tma_store_bound)) * (tma_lock_latency / (tma_= dtlb_load + tma_store_fwd_blk + tma_l1_latency_dependency + tma_lock_latenc= y + tma_split_loads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * = (tma_l1_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_boun= d + tma_store_bound)) * (tma_split_loads / (tma_dtlb_load + tma_store_fwd_b= lk + tma_l1_latency_dependency + tma_lock_latency + tma_split_loads + tma_4= k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound / (tma_l1_= bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound)) * = (tma_split_stores / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store)) + tma_memory_bound * (tma_stor= e_bound / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tm= a_store_bound)) * (tma_store_latency / (tma_store_latency + tma_false_shari= ng + tma_split_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_dram_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) *= (tma_mem_latency / (tma_mem_bandwidth + tma_mem_latency)) + tma_memory_bou= nd * (tma_l3_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3= _bound + tma_store_bound)) * (tma_l3_hit_latency / (tma_contested_accesses = + tma_data_sharing + tma_l3_hit_latency + tma_sq_full)) + tma_memory_bound = * tma_l2_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bou= nd + tma_store_bound) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound = + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_l1_= latency_dependency / (tma_4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l= 1_latency_dependency + tma_lock_latency + tma_split_loads + tma_store_fwd_b= lk)) + tma_memory_bound * (tma_l1_bound / (tma_dram_bound + tma_l1_bound + = tma_l2_bound + tma_l3_bound + tma_store_bound)) * (tma_lock_latency / (tma_= 4k_aliasing + tma_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma= _lock_latency + tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * = (tma_l1_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_boun= d + tma_store_bound)) * (tma_split_loads / (tma_4k_aliasing + tma_dtlb_load= + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + tma_split_l= oads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound / (tma_dra= m_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound)) * = (tma_split_stores / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores)) + tma_memory_bound * (tma_stor= e_bound / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tm= a_store_bound)) * (tma_store_latency / (tma_dtlb_store + tma_false_sharing = + tma_split_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvML;Mem;MemoryLat;Offcore;tma_issueLat", "MetricName": "tma_bottleneck_cache_memory_latency", "MetricThreshold": "tma_bottleneck_cache_memory_latency > 20", @@ -165,22 +165,22 @@ }, { "BriefDescription": "Total pipeline cost when the execution is com= pute-bound - an estimation", - "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_serializing_operation + tma_ports_utilization) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_serializing_operation + tma_ports_= utilization)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", + "MetricExpr": "100 * (tma_core_bound * tma_divider / (tma_divider = + tma_ports_utilization + tma_serializing_operation) + tma_core_bound * (tm= a_ports_utilization / (tma_divider + tma_ports_utilization + tma_serializin= g_operation)) * (tma_ports_utilized_3m / (tma_ports_utilized_0 + tma_ports_= utilized_1 + tma_ports_utilized_2 + tma_ports_utilized_3m)))", "MetricGroup": "BvCB;Cor;tma_issueComp", "MetricName": "tma_bottleneck_compute_bound_est", "MetricThreshold": "tma_bottleneck_compute_bound_est > 20", - "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy" + "PublicDescription": "Total pipeline cost when the execution is co= mpute-bound - an estimation. Covers Core Bound when High ILP as well as whe= n long-latency execution units are busy. Related metrics: " }, { "BriefDescription": "Total pipeline cost of instruction fetch band= width related bottlenecks (when the front-end could not sustain operations = delivery to the back-end)", - "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses + t= ma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_mispredicts_resteers + tma_clears_resteers + tma_unknown_br= anches)) / (tma_icache_misses + tma_itlb_misses + tma_branch_resteers + tma= _ms_switches + tma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_ms /= (tma_mite + tma_dsb + tma_lsd + tma_ms))) - tma_bottleneck_big_code", + "MetricExpr": "100 * (tma_frontend_bound - (1 - 10 * tma_microcode= _sequencer * tma_other_mispredicts / tma_branch_mispredicts) * tma_fetch_la= tency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switches = + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches) - tma_mi= crocode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer) *= (tma_assists / tma_microcode_sequencer) * (tma_fetch_latency * (tma_ms_swi= tches + tma_branch_resteers * (tma_clears_resteers + 10 * tma_microcode_seq= uencer * tma_other_mispredicts / tma_branch_mispredicts * tma_mispredicts_r= esteers) / (tma_clears_resteers + tma_mispredicts_resteers + tma_unknown_br= anches)) / (tma_branch_resteers + tma_dsb_switches + tma_icache_misses + tm= a_itlb_misses + tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_ms /= (tma_dsb + tma_lsd + tma_mite + tma_ms))) - tma_bottleneck_big_code", "MetricGroup": "BvFB;Fed;FetchBW;Frontend", "MetricName": "tma_bottleneck_instruction_fetch_bw", "MetricThreshold": "tma_bottleneck_instruction_fetch_bw > 20" }, { "BriefDescription": "Total pipeline cost of irregular execution (e= .g", - "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_mispredicts_resteers += tma_clears_resteers + tma_unknown_branches)) / (tma_icache_misses + tma_it= lb_misses + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_mite + tma_dsb + tma_lsd + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_serializing_operation + tma_ports_utilization) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", + "MetricExpr": "100 * (tma_microcode_sequencer / (tma_few_uops_inst= ructions + tma_microcode_sequencer) * (tma_assists / tma_microcode_sequence= r) * (tma_fetch_latency * (tma_ms_switches + tma_branch_resteers * (tma_cle= ars_resteers + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_b= ranch_mispredicts * tma_mispredicts_resteers) / (tma_clears_resteers + tma_= mispredicts_resteers + tma_unknown_branches)) / (tma_branch_resteers + tma_= dsb_switches + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switc= hes) + tma_fetch_bandwidth * tma_ms / (tma_dsb + tma_lsd + tma_mite + tma_m= s)) + 10 * tma_microcode_sequencer * tma_other_mispredicts / tma_branch_mis= predicts * tma_branch_mispredicts + tma_machine_clears * tma_other_nukes / = tma_other_nukes + tma_core_bound * (tma_serializing_operation + tma_core_bo= und * RS_EVENTS.EMPTY_CYCLES / tma_info_thread_clks * tma_ports_utilized_0)= / (tma_divider + tma_ports_utilization + tma_serializing_operation) + tma_= microcode_sequencer / (tma_few_uops_instructions + tma_microcode_sequencer)= * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "Bad;BvIO;Cor;Ret;tma_issueMS", "MetricName": "tma_bottleneck_irregular_overhead", "MetricThreshold": "tma_bottleneck_irregular_overhead > 10", @@ -188,7 +188,7 @@ }, { "BriefDescription": "Total pipeline cost of Memory Address Transla= tion related bottlenecks (data-side TLBs)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_dtlb_load + tma_= store_fwd_blk + tma_l1_latency_dependency + tma_lock_latency + tma_split_lo= ads + tma_4k_aliasing + tma_fb_full)) + tma_memory_bound * (tma_store_bound= / (tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_store_latency + tma_false_sharing + tma_s= plit_stores + tma_streaming_stores + tma_dtlb_store)))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l1_bound / max(tma_m= emory_bound, tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + = tma_store_bound)) * (tma_dtlb_load / max(tma_l1_bound, tma_4k_aliasing + tm= a_dtlb_load + tma_fb_full + tma_l1_latency_dependency + tma_lock_latency + = tma_split_loads + tma_store_fwd_blk)) + tma_memory_bound * (tma_store_bound= / (tma_dram_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store= _bound)) * (tma_dtlb_store / (tma_dtlb_store + tma_false_sharing + tma_spli= t_stores + tma_store_latency + tma_streaming_stores)))", "MetricGroup": "BvMT;Mem;MemoryTLB;Offcore;tma_issueTLB", "MetricName": "tma_bottleneck_memory_data_tlbs", "MetricThreshold": "tma_bottleneck_memory_data_tlbs > 20", @@ -196,15 +196,15 @@ }, { "BriefDescription": "Total pipeline cost of Memory Synchronization= related bottlenecks (data transfers and coherency updates across processor= s)", - "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_l1_b= ound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_l1= _bound + tma_l2_bound + tma_l3_bound + tma_dram_bound + tma_store_bound) * = tma_false_sharing / (tma_store_latency + tma_false_sharing + tma_split_stor= es + tma_streaming_stores + tma_dtlb_store - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", + "MetricExpr": "100 * (tma_memory_bound * (tma_l3_bound / (tma_dram= _bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * (t= ma_contested_accesses + tma_data_sharing) / (tma_contested_accesses + tma_d= ata_sharing + tma_l3_hit_latency + tma_sq_full) + tma_store_bound / (tma_dr= am_bound + tma_l1_bound + tma_l2_bound + tma_l3_bound + tma_store_bound) * = tma_false_sharing / (tma_dtlb_store + tma_false_sharing + tma_split_stores = + tma_store_latency + tma_streaming_stores - tma_store_latency)) + tma_mach= ine_clears * (1 - tma_other_nukes / tma_other_nukes))", "MetricGroup": "BvMS;LockCont;Mem;Offcore;tma_issueSyncxn", "MetricName": "tma_bottleneck_memory_synchronization", "MetricThreshold": "tma_bottleneck_memory_synchronization > 10", - "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears" + "PublicDescription": "Total pipeline cost of Memory Synchronizatio= n related bottlenecks (data transfers and coherency updates across processo= rs). Related metrics: tma_contested_accesses, tma_data_sharing, tma_false_s= haring, tma_machine_clears, tma_remote_cache" }, { "BriefDescription": "Total pipeline cost of Branch Misprediction r= elated bottlenecks", - "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_icache_misses + tma_itlb_misses= + tma_branch_resteers + tma_ms_switches + tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (1 - 10 * tma_microcode_sequencer * tma_other= _mispredicts / tma_branch_mispredicts) * (tma_branch_mispredicts + tma_fetc= h_latency * tma_mispredicts_resteers / (tma_branch_resteers + tma_dsb_switc= hes + tma_icache_misses + tma_itlb_misses + tma_lcp + tma_ms_switches))", "MetricGroup": "Bad;BadSpec;BrMispredicts;BvMP;tma_issueBM", "MetricName": "tma_bottleneck_mispredictions", "MetricThreshold": "tma_bottleneck_mispredictions > 20", @@ -216,17 +216,17 @@ "MetricGroup": "BvOB;Cor;Offcore", "MetricName": "tma_bottleneck_other_bottlenecks", "MetricThreshold": "tma_bottleneck_other_bottlenecks > 20", - "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls" + "PublicDescription": "Total pipeline cost of remaining bottlenecks= in the back-end. Examples include data-dependencies (Core Bound when Low I= LP) and other unlisted memory-related stalls." }, { - "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead", + "BriefDescription": "Total pipeline cost of \"useful operations\" = - the portion of Retiring category not covered by Branching_Overhead nor Ir= regular_Overhead.", "MetricExpr": "100 * (tma_retiring - (BR_INST_RETIRED.ALL_BRANCHES= + 2 * BR_INST_RETIRED.NEAR_CALL + INST_RETIRED.NOP) / tma_info_thread_slot= s - tma_microcode_sequencer / (tma_few_uops_instructions + tma_microcode_se= quencer) * (tma_assists / tma_microcode_sequencer) * tma_heavy_operations)", "MetricGroup": "BvUW;Ret", "MetricName": "tma_bottleneck_useful_work", "MetricThreshold": "tma_bottleneck_useful_work > 20" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring branch instructions.", "MetricExpr": "tma_light_operations * BR_INST_RETIRED.ALL_BRANCHES= / (tma_retiring * tma_info_thread_slots)", "MetricGroup": "Branches;BvBO;Pipeline;TopdownL3;tma_L3_group;tma_= light_operations_group", "MetricName": "tma_branch_instructions", @@ -248,8 +248,8 @@ "MetricExpr": "INT_MISC.CLEAR_RESTEER_CYCLES / tma_info_thread_clk= s + tma_unknown_branches", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group", "MetricName": "tma_branch_resteers", - "MetricThreshold": "tma_branch_resteers > 0.05 & tma_fetch_latency= > 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES. Related metrics: tma_l3_hit_latency, tma_store_latency", + "MetricThreshold": "tma_branch_resteers > 0.05 & (tma_fetch_latenc= y > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers. Branch Resteers estimates the Fro= ntend delay in fetching operations from corrected path; following all sorts= of miss-predicted branches. For example; branchy code with lots of miss-pr= edictions might get categorized under Branch Resteers. Note the value of th= is node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRA= NCHES", "ScaleUnit": "100%" }, { @@ -257,8 +257,8 @@ "MetricExpr": "max(0, tma_microcode_sequencer - tma_assists)", "MetricGroup": "TopdownL4;tma_L4_group;tma_microcode_sequencer_gro= up", "MetricName": "tma_cisc", - "MetricThreshold": "tma_cisc > 0.1 & tma_microcode_sequencer > 0.0= 5 & tma_heavy_operations > 0.1", - "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources", + "MetricThreshold": "tma_cisc > 0.1 & (tma_microcode_sequencer > 0.= 05 & tma_heavy_operations > 0.1)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU retired uops originated from CISC (complex instruction set computer) i= nstruction. A CISC instruction has multiple uops that are required to perfo= rm the instruction's functionality as in the case of read-modify-write as a= n example. Since these instructions require multiple uops they may or may n= ot imply sub-optimal use of machine resources.", "ScaleUnit": "100%" }, { @@ -266,24 +266,24 @@ "MetricExpr": "(1 - BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRE= D.ALL_BRANCHES + MACHINE_CLEARS.COUNT)) * INT_MISC.CLEAR_RESTEER_CYCLES / t= ma_info_thread_clks", "MetricGroup": "BadSpec;MachineClears;TopdownL4;tma_L4_group;tma_b= ranch_resteers_group;tma_issueMC", "MetricName": "tma_clears_resteers", - "MetricThreshold": "tma_clears_resteers > 0.05 & tma_branch_restee= rs > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_clears_resteers > 0.05 & (tma_branch_reste= ers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Machine Clears. Sam= ple with: INT_MISC.CLEAR_RESTEER_CYCLES. Related metrics: tma_l1_bound, tma= _machine_clears, tma_microcode_sequencer, tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that hit in the L2 cache.", "MetricExpr": "max(0, tma_icache_misses - tma_code_l2_miss)", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_hit", - "MetricThreshold": "tma_code_l2_hit > 0.05 & tma_icache_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_hit > 0.05 & (tma_icache_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache", + "BriefDescription": "This metric estimates fraction of cycles the = CPU was stalled due to instruction cache misses that miss in the L2 cache.", "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.CYCLES_WITH_DEMAND_COD= E_RD / tma_info_thread_clks", "MetricGroup": "FetchLat;IcMiss;Offcore;TopdownL4;tma_L4_group;tma= _icache_misses_group", "MetricName": "tma_code_l2_miss", - "MetricThreshold": "tma_code_l2_miss > 0.05 & tma_icache_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_l2_miss > 0.05 & (tma_icache_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -291,7 +291,7 @@ "MetricExpr": "max(0, tma_itlb_misses - tma_code_stlb_miss)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_hit", - "MetricThreshold": "tma_code_stlb_hit > 0.05 & tma_itlb_misses > 0= .05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_hit > 0.05 & (tma_itlb_misses > = 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { @@ -299,33 +299,33 @@ "MetricExpr": "ITLB_MISSES.WALK_ACTIVE / tma_info_thread_clks", "MetricGroup": "FetchLat;MemoryTLB;TopdownL4;tma_L4_group;tma_itlb= _misses_group", "MetricName": "tma_code_stlb_miss", - "MetricThreshold": "tma_code_stlb_miss > 0.05 & tma_itlb_misses > = 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss > 0.05 & (tma_itlb_misses >= 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_2M_= 4M / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_2m", - "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_2m > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= (instruction) code accesses.", "MetricExpr": "tma_code_stlb_miss * ITLB_MISSES.WALK_COMPLETED_4K = / (ITLB_MISSES.WALK_COMPLETED_4K + ITLB_MISSES.WALK_COMPLETED_2M_4M)", "MetricGroup": "FetchLat;MemoryTLB;TopdownL5;tma_L5_group;tma_code= _stlb_miss_group", "MetricName": "tma_code_stlb_miss_4k", - "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & tma_code_stlb_m= iss > 0.05 & tma_itlb_misses > 0.05 & tma_fetch_latency > 0.1 & tma_fronten= d_bound > 0.15", + "MetricThreshold": "tma_code_stlb_miss_4k > 0.05 & (tma_code_stlb_= miss > 0.05 & (tma_itlb_misses > 0.05 & (tma_fetch_latency > 0.1 & tma_fron= tend_bound > 0.15)))", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to contested acces= ses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "((54 * tma_info_system_core_frequency - 5 * tma_inf= o_system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (OCR.DEMAND_= DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEM= AND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) + (53 * tma_info_system_core_frequ= ency - 5 * tma_info_system_core_frequency) * MEM_LOAD_L3_HIT_RETIRED.XSNP_M= ISS) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_i= nfo_thread_clks", + "MetricExpr": "(49 * tma_info_system_core_frequency * (MEM_LOAD_L3= _HIT_RETIRED.XSNP_FWD * (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND= _DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))= ) + 48 * tma_info_system_core_frequency * MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS= ) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info= _thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_l3_bound_group", "MetricName": "tma_contested_accesses", - "MetricThreshold": "tma_contested_accesses > 0.05 & tma_l3_bound >= 0.05 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD, MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related = metrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false= _sharing, tma_machine_clears", + "MetricThreshold": "tma_contested_accesses > 0.05 & (tma_l3_bound = > 0.05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to contested acce= sses. Contested accesses occur when data written by one Logical Processor a= re read by another Logical Processor on a different Physical Core. Examples= of contested accesses include synchronizations such as locks; true data sh= aring such as modified locked variables; and false sharing. Sample with: ME= M_LOAD_L3_HIT_RETIRED.XSNP_FWD;MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS. Related m= etrics: tma_bottleneck_memory_synchronization, tma_data_sharing, tma_false_= sharing, tma_machine_clears, tma_remote_cache", "ScaleUnit": "100%" }, { @@ -335,25 +335,25 @@ "MetricName": "tma_core_bound", "MetricThreshold": "tma_core_bound > 0.1 & tma_backend_bound > 0.2= ", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations)", + "PublicDescription": "This metric represents fraction of slots whe= re Core non-memory issues were of a bottleneck. Shortage in hardware compu= te resources; or dependencies in software's instructions are both categoriz= ed under Core Bound. Hence it may indicate the machine ran out of an out-of= -order resource; certain execution units are overloaded or dependencies in = program's data- or instruction-flow are limiting the performance (e.g. FP-c= hained long-latency arithmetic operations).", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles whil= e the memory subsystem was handling synchronizations due to data-sharing ac= cesses", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "(53 * tma_info_system_core_frequency - 5 * tma_info= _system_core_frequency) * (MEM_LOAD_L3_HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L= 3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_= FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / tma= _info_thread_clks", + "MetricExpr": "48 * tma_info_system_core_frequency * (MEM_LOAD_L3_= HIT_RETIRED.XSNP_NO_FWD + MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD * (1 - OCR.DEMAN= D_DATA_RD.L3_HIT.SNOOP_HITM / (OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM + OCR.D= EMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD))) * (1 + MEM_LOAD_RETIRED.FB_HIT /= MEM_LOAD_RETIRED.L1_MISS / 2) / tma_info_thread_clks", "MetricGroup": "BvMS;Offcore;Snoop;TopdownL4;tma_L4_group;tma_issu= eSyncxn;tma_l3_bound_group", "MetricName": "tma_data_sharing", - "MetricThreshold": "tma_data_sharing > 0.05 & tma_l3_bound > 0.05 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears", + "MetricThreshold": "tma_data_sharing > 0.05 & (tma_l3_bound > 0.05= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles whi= le the memory subsystem was handling synchronizations due to data-sharing a= ccesses. Data shared by multiple Logical Processors (even just read shared)= may cause increased access latency due to cache coherency. Excessive data = sharing can drastically harm multithreaded performance. Sample with: MEM_LO= AD_L3_HIT_RETIRED.XSNP_NO_FWD. Related metrics: tma_bottleneck_memory_synch= ronization, tma_contested_accesses, tma_false_sharing, tma_machine_clears, = tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric represents fraction of cycles whe= re decoder-0 was the only active decoder", - "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D0x1@ - cpu@I= NST_DECODED.DECODERS\\,cmask\\=3D0x2@) / tma_info_core_core_clks / 2", + "MetricExpr": "(cpu@INST_DECODED.DECODERS\\,cmask\\=3D1@ - cpu@INS= T_DECODED.DECODERS\\,cmask\\=3D2@) / tma_info_core_core_clks / 2", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_issueD0= ;tma_mite_group", "MetricName": "tma_decoder0_alone", - "MetricThreshold": "tma_decoder0_alone > 0.1 & tma_mite > 0.1 & tm= a_fetch_bandwidth > 0.2", + "MetricThreshold": "tma_decoder0_alone > 0.1 & (tma_mite > 0.1 & t= ma_fetch_bandwidth > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere decoder-0 was the only active decoder. Related metrics: tma_few_uops_in= structions", "ScaleUnit": "100%" }, @@ -362,7 +362,7 @@ "MetricExpr": "ARITH.DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "BvCB;TopdownL3;tma_L3_group;tma_core_bound_group", "MetricName": "tma_divider", - "MetricThreshold": "tma_divider > 0.2 & tma_core_bound > 0.1 & tma= _backend_bound > 0.2", + "MetricThreshold": "tma_divider > 0.2 & (tma_core_bound > 0.1 & tm= a_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles wh= ere the Divider unit was active. Divide and square root instructions are pe= rformed by the Divider unit and can take considerably longer latency than i= nteger or Floating Point addition; subtraction; or multiplication. Sample w= ith: ARITH.DIVIDER_ACTIVE", "ScaleUnit": "100%" }, @@ -372,7 +372,7 @@ "MetricExpr": "CYCLE_ACTIVITY.STALLS_L3_MISS / tma_info_thread_clk= s + (CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks - tma_l2_bound", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_dram_bound", - "MetricThreshold": "tma_dram_bound > 0.1 & tma_memory_bound > 0.2 = & tma_backend_bound > 0.2", + "MetricThreshold": "tma_dram_bound > 0.1 & (tma_memory_bound > 0.2= & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled on accesses to external memory (DRAM) by loads. Better caching can = improve the latency and increase performance. Sample with: MEM_LOAD_RETIRED= .L3_MISS", "ScaleUnit": "100%" }, @@ -382,7 +382,7 @@ "MetricGroup": "DSB;FetchBW;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_dsb", "MetricThreshold": "tma_dsb > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to DSB (decoded uop cache) fetch pip= eline. For example; inefficient utilization of the DSB cache structure or = bank conflict when reading from it; are categorized here.", "ScaleUnit": "100%" }, { @@ -390,26 +390,26 @@ "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / tma_info_thread_= clks", "MetricGroup": "DSBmiss;FetchLat;TopdownL3;tma_L3_group;tma_fetch_= latency_group;tma_issueFB", "MetricName": "tma_dsb_switches", - "MetricThreshold": "tma_dsb_switches > 0.05 & tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwi= dth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_inf= o_inst_mix_iptb, tma_lcp", + "MetricThreshold": "tma_dsb_switches > 0.05 & (tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to switches from DSB to MITE pipelines. The DSB (deco= ded i-cache) is a Uop Cache where the front-end directly delivers Uops (mic= ro operations) avoiding heavy x86 decoding. The DSB pipeline has shorter la= tency and delivered higher bandwidth than the MITE (legacy instruction deco= de pipeline). Switching between the two pipelines can cause penalties hence= this metric measures the exposed penalty. Sample with: FRONTEND_RETIRED.DS= B_MISS_PS. Related metrics: tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_ban= dwidth, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_= info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles where the Data TLB (DTLB) was missed by load accesses", - "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D0= x1@ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYC= LE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", + "MetricExpr": "min(7 * cpu@DTLB_LOAD_MISSES.STLB_HIT\\,cmask\\=3D1= @ + DTLB_LOAD_MISSES.WALK_ACTIVE, max(CYCLE_ACTIVITY.CYCLES_MEM_ANY - CYCLE= _ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_l1_bound_group", "MetricName": "tma_dtlb_load", - "MetricThreshold": "tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma= _memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS. Related metrics: tma_= bottleneck_memory_data_tlbs, tma_dtlb_store", + "MetricThreshold": "tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (t= ma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles where the Data TLB (DTLB) was missed by load accesses. TLBs (Trans= lation Look-aside Buffers) are processor caches for recently used entries o= ut of the Page Tables that are used to map virtual- to physical-addresses b= y the operating system. This metric approximates the potential delay of dem= and loads missing the first-level data TLB (assuming worst case scenario wi= th back to back misses to different pages). This includes hitting in the se= cond-level TLB (STLB) as well as performing a hardware page walk on an STLB= miss. Sample with: MEM_INST_RETIRED.STLB_MISS_LOADS_PS. Related metrics: t= ma_bottleneck_memory_data_tlbs, tma_dtlb_store", "ScaleUnit": "100%" }, { "BriefDescription": "This metric roughly estimates the fraction of= cycles spent handling first-level data TLB store misses", - "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D0x1= @ + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", + "MetricExpr": "(7 * cpu@DTLB_STORE_MISSES.STLB_HIT\\,cmask\\=3D1@ = + DTLB_STORE_MISSES.WALK_ACTIVE) / tma_info_core_core_clks", "MetricGroup": "BvMT;MemoryTLB;TopdownL4;tma_L4_group;tma_issueTLB= ;tma_store_bound_group", "MetricName": "tma_dtlb_store", - "MetricThreshold": "tma_dtlb_store > 0.05 & tma_store_bound > 0.2 = & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES. Related metrics: tma_bottleneck_mem= ory_data_tlbs, tma_dtlb_load", + "MetricThreshold": "tma_dtlb_store > 0.05 & (tma_store_bound > 0.2= & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates the fraction o= f cycles spent handling first-level data TLB store misses. As with ordinar= y data caching; focus on improving data locality and reducing working-set s= ize to reduce DTLB overhead. Additionally; consider using profile-guided o= ptimization (PGO) to collocate frequently-used data on the same page. Try = using larger page sizes for large amounts of frequently-used data. Sample w= ith: MEM_INST_RETIRED.STLB_MISS_STORES_PS. Related metrics: tma_bottleneck_= memory_data_tlbs, tma_dtlb_load", "ScaleUnit": "100%" }, { @@ -417,8 +417,8 @@ "MetricExpr": "54 * tma_info_system_core_frequency * OCR.DEMAND_RF= O.L3_HIT.SNOOP_HITM / tma_info_thread_clks", "MetricGroup": "BvMS;DataSharing;LockCont;Offcore;Snoop;TopdownL4;= tma_L4_group;tma_issueSyncxn;tma_store_bound_group", "MetricName": "tma_false_sharing", - "MetricThreshold": "tma_false_sharing > 0.05 & tma_store_bound > 0= .2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears", + "MetricThreshold": "tma_false_sharing > 0.05 & (tma_store_bound > = 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates how often CPU = was handling synchronizations due to False Sharing. False Sharing is a mult= ithreading hiccup; where multiple Logical Processors contend on different d= ata-elements mapped into the same cache line. Sample with: OCR.DEMAND_RFO.L= 3_HIT.SNOOP_HITM. Related metrics: tma_bottleneck_memory_synchronization, t= ma_contested_accesses, tma_data_sharing, tma_machine_clears, tma_remote_cac= he", "ScaleUnit": "100%" }, { @@ -437,7 +437,7 @@ "MetricName": "tma_fetch_bandwidth", "MetricThreshold": "tma_fetch_bandwidth > 0.2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1, FRONTEND_RETIRED.LATE= NCY_GE_1, FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches,= tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_= frontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend bandwidth issues. For example; inefficien= cies at the instruction decoders; or restrictions for caching in the DSB (d= ecoded uops cache) are categorized under Fetch Bandwidth. In such cases; th= e Frontend typically delivers suboptimal amount of uops to the Backend. Sam= ple with: FRONTEND_RETIRED.LATENCY_GE_2_BUBBLES_GE_1;FRONTEND_RETIRED.LATEN= CY_GE_1;FRONTEND_RETIRED.LATENCY_GE_2. Related metrics: tma_dsb_switches, t= ma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses, tma_info_fr= ontend_dsb_coverage, tma_info_inst_mix_iptb, tma_lcp", "ScaleUnit": "100%" }, { @@ -447,7 +447,7 @@ "MetricName": "tma_fetch_latency", "MetricThreshold": "tma_fetch_latency > 0.1 & tma_frontend_bound >= 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16, FRONTEND_RETIRED.LATENCY_GE_8", + "PublicDescription": "This metric represents fraction of slots the= CPU was stalled due to Frontend latency issues. For example; instruction-= cache misses; iTLB misses or fetch stalls after a branch misprediction are = categorized under Frontend Latency. In such cases; the Frontend eventually = delivers no uops for some period. Sample with: FRONTEND_RETIRED.LATENCY_GE_= 16_PS;FRONTEND_RETIRED.LATENCY_GE_8_PS", "ScaleUnit": "100%" }, { @@ -465,7 +465,7 @@ "MetricGroup": "HPC;TopdownL3;tma_L3_group;tma_light_operations_gr= oup", "MetricName": "tma_fp_arith", "MetricThreshold": "tma_fp_arith > 0.2 & tma_light_operations > 0.= 6", - "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting", + "PublicDescription": "This metric represents overall arithmetic fl= oating-point (FP) operations fraction the CPU has executed (retired). Note = this metric's value may exceed its parent due to use of \"Uops\" CountDomai= n and FMA double-counting.", "ScaleUnit": "100%" }, { @@ -474,15 +474,15 @@ "MetricGroup": "HPC;TopdownL5;tma_L5_group;tma_assists_group", "MetricName": "tma_fp_assists", "MetricThreshold": "tma_fp_assists > 0.1", - "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals)", + "PublicDescription": "This metric roughly estimates fraction of sl= ots the CPU retired uops as a result of handing Floating Point (FP) Assists= . FP Assist may apply when working with very small floating point values (s= o-called Denormals).", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Floating-Point Divider unit was active.", "MetricExpr": "ARITH.FP_DIVIDER_ACTIVE / tma_info_thread_clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_fp_divider", - "MetricThreshold": "tma_fp_divider > 0.2 & tma_divider > 0.2 & tma= _core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_fp_divider > 0.2 & (tma_divider > 0.2 & (t= ma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -490,7 +490,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.SCALAR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_scalar", - "MetricThreshold": "tma_fp_scalar > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_scalar > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) scalar uops fraction the CPU has retired. May overcount due to = FMA double counting. Related metrics: tma_fp_vector, tma_fp_vector_128b, tm= a_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_1, tma_port_5, t= ma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -499,7 +499,7 @@ "MetricExpr": "FP_ARITH_INST_RETIRED.VECTOR / (tma_retiring * tma_= info_thread_slots)", "MetricGroup": "Compute;Flops;TopdownL4;tma_L4_group;tma_fp_arith_= group;tma_issue2P", "MetricName": "tma_fp_vector", - "MetricThreshold": "tma_fp_vector > 0.1 & tma_fp_arith > 0.2 & tma= _light_operations > 0.6", + "MetricThreshold": "tma_fp_vector > 0.1 & (tma_fp_arith > 0.2 & tm= a_light_operations > 0.6)", "PublicDescription": "This metric approximates arithmetic floating= -point (FP) vector uops fraction the CPU has retired aggregated across all = vector widths. May overcount due to FMA double counting. Related metrics: t= ma_fp_scalar, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, t= ma_port_0, tma_port_1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -508,7 +508,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.128B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_128b", - "MetricThreshold": "tma_fp_vector_128b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_128b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 128-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_256b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -517,7 +517,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.256B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_256b", - "MetricThreshold": "tma_fp_vector_256b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_256b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 256-bit wide vectors. May overcount= due to FMA double counting prior to LNL. Related metrics: tma_fp_scalar, t= ma_fp_vector, tma_fp_vector_128b, tma_fp_vector_512b, tma_port_0, tma_port_= 1, tma_port_5, tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -526,7 +526,7 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARIT= H_INST_RETIRED.512B_PACKED_SINGLE) / (tma_retiring * tma_info_thread_slots)= ", "MetricGroup": "Compute;Flops;TopdownL5;tma_L5_group;tma_fp_vector= _group;tma_issue2P", "MetricName": "tma_fp_vector_512b", - "MetricThreshold": "tma_fp_vector_512b > 0.1 & tma_fp_vector > 0.1= & tma_fp_arith > 0.2 & tma_light_operations > 0.6", + "MetricThreshold": "tma_fp_vector_512b > 0.1 & (tma_fp_vector > 0.= 1 & (tma_fp_arith > 0.2 & tma_light_operations > 0.6))", "PublicDescription": "This metric approximates arithmetic FP vecto= r uops fraction the CPU has retired for 512-bit wide vectors. May overcount= due to FMA double counting. Related metrics: tma_fp_scalar, tma_fp_vector,= tma_fp_vector_128b, tma_fp_vector_256b, tma_port_0, tma_port_1, tma_port_5= , tma_port_6, tma_ports_utilized_2", "ScaleUnit": "100%" }, @@ -538,17 +538,17 @@ "MetricName": "tma_frontend_bound", "MetricThreshold": "tma_frontend_bound > 0.15", "MetricgroupNoGroup": "TopdownL1;Default", - "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4", + "PublicDescription": "This category represents fraction of slots w= here the processor's Frontend undersupplies its Backend. Frontend denotes t= he first part of the processor core responsible to fetch operations that ar= e executed later on by the Backend part. Within the Frontend; a branch pred= ictor predicts the next address to fetch; cache-lines are fetched from the = memory subsystem; parsed into instructions; and lastly decoded into micro-o= perations (uops). Ideally the Frontend can issue Pipeline_Width uops every = cycle to the Backend. Frontend Bound denotes unutilized issue-slots when th= ere is no Backend stall; i.e. bubbles where Frontend delivered no uops whil= e Backend could have accepted them. For example; stalls due to instruction-= cache misses would be categorized under Frontend Bound. Sample with: FRONTE= ND_RETIRED.LATENCY_GE_4_PS", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations , instructions that require = two or more uops or micro-coded sequences", - "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D0x1@) / IDQ.MITE_UOPS", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring heavy-weight operations -- instructions that require= two or more uops or micro-coded sequences", + "MetricExpr": "tma_microcode_sequencer + tma_retiring * (UOPS_DECO= DED.DEC0 - cpu@UOPS_DECODED.DEC0\\,cmask\\=3D1@) / IDQ.MITE_UOPS", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_heavy_operations", "MetricThreshold": "tma_heavy_operations > 0.1", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations , instructions that require= two or more uops or micro-coded sequences. This highly-correlates with the= uop length of these instructions/sequences.([ICL+] Note this may overcount= due to approximation using indirect events; [ADL+])", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring heavy-weight operations -- instructions that requir= e two or more uops or micro-coded sequences. This highly-correlates with th= e uop length of these instructions/sequences.([ICL+] Note this may overcoun= t due to approximation using indirect events; [ADL+])", "ScaleUnit": "100%" }, { @@ -556,8 +556,8 @@ "MetricExpr": "ICACHE_DATA.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;IcMiss;TopdownL3;tma_L3= _group;tma_fetch_latency_group", "MetricName": "tma_icache_misses", - "MetricThreshold": "tma_icache_misses > 0.05 & tma_fetch_latency >= 0.1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS, FRONTEND_RETIRED.L1I_MISS", + "MetricThreshold": "tma_icache_misses > 0.05 & (tma_fetch_latency = > 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to instruction cache misses. Sample with: FRONTEND_RE= TIRED.L2_MISS_PS;FRONTEND_RETIRED.L1I_MISS_PS", "ScaleUnit": "100%" }, { @@ -569,28 +569,28 @@ "PublicDescription": "Branch Misprediction Cost: Cycles representi= ng fraction of TMA slots wasted per non-speculative branch misprediction (r= etired JEClear). Related metrics: tma_bottleneck_mispredictions, tma_branch= _mispredicts, tma_mispredicts_resteers" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional non-taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_NTAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_ntaken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_ntaken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for cond= itional taken branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.COND_TAKEN", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_cond_taken", "MetricThreshold": "tma_info_bad_spec_ipmisp_cond_taken < 200" }, { - "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for indi= rect CALL or JMP branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.INDIRECT", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_indirect", - "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1000" + "MetricThreshold": "tma_info_bad_spec_ipmisp_indirect < 1e3" }, { - "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate)", + "BriefDescription": "Instructions per retired Mispredicts for retu= rn branches (lower number means higher occurrence rate).", "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.RET", "MetricGroup": "Bad;BrMispredicts", "MetricName": "tma_info_bad_spec_ipmisp_ret", @@ -619,7 +619,7 @@ }, { "BriefDescription": "Total pipeline cost of DSB (uop cache) hits -= subset of the Instruction_Fetch_BW Bottleneck", - "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_latency + tma_fetch_bandwidth)) * (tma_dsb / (tma_mite + tma_dsb= + tma_lsd + tma_ms)))", + "MetricExpr": "100 * (tma_frontend_bound * (tma_fetch_bandwidth / = (tma_fetch_bandwidth + tma_fetch_latency)) * (tma_dsb / (tma_dsb + tma_lsd = + tma_mite + tma_ms)))", "MetricGroup": "DSB;Fed;FetchBW;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_bandwidth", "MetricThreshold": "tma_info_botlnk_l2_dsb_bandwidth > 10", @@ -628,7 +628,7 @@ { "BriefDescription": "Total pipeline cost of DSB (uop cache) misses= - subset of the Instruction_Fetch_BW Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + t= ma_lcp + tma_dsb_switches) + tma_fetch_bandwidth * tma_mite / (tma_mite + t= ma_dsb + tma_lsd + tma_ms))", + "MetricExpr": "100 * (tma_fetch_latency * tma_dsb_switches / (tma_= branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses + = tma_lcp + tma_ms_switches) + tma_fetch_bandwidth * tma_mite / (tma_dsb + tm= a_lsd + tma_mite + tma_ms))", "MetricGroup": "DSBmiss;Fed;tma_issueFB", "MetricName": "tma_info_botlnk_l2_dsb_misses", "MetricThreshold": "tma_info_botlnk_l2_dsb_misses > 10", @@ -637,10 +637,11 @@ { "BriefDescription": "Total pipeline cost of Instruction Cache miss= es - subset of the Big_Code Bottleneck", "MetricConstraint": "NO_GROUP_EVENTS", - "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _icache_misses + tma_itlb_misses + tma_branch_resteers + tma_ms_switches + = tma_lcp + tma_dsb_switches))", + "MetricExpr": "100 * (tma_fetch_latency * tma_icache_misses / (tma= _branch_resteers + tma_dsb_switches + tma_icache_misses + tma_itlb_misses += tma_lcp + tma_ms_switches))", "MetricGroup": "Fed;FetchLat;IcMiss;tma_issueFL", "MetricName": "tma_info_botlnk_l2_ic_misses", - "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5" + "MetricThreshold": "tma_info_botlnk_l2_ic_misses > 5", + "PublicDescription": "Total pipeline cost of Instruction Cache mis= ses - subset of the Big_Code Bottleneck. Related metrics: " }, { "BriefDescription": "Fraction of branches that are CALL or RET", @@ -701,11 +702,11 @@ "MetricExpr": "(FP_ARITH_INST_RETIRED.SCALAR + FP_ARITH_INST_RETIR= ED.VECTOR) / (2 * tma_info_core_core_clks)", "MetricGroup": "Cor;Flops;HPC", "MetricName": "tma_info_core_fp_arith_utilization", - "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)" + "PublicDescription": "Actual per-core usage of the Floating Point = non-X87 execution units (regardless of precision or vector-width). Values >= 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; = [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less commo= n)." }, { "BriefDescription": "Instruction-Level-Parallelism (average number= of uops executed when there is execution) per thread (logical-processor)", - "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D0x1@", + "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,c= mask\\=3D1@", "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", "MetricName": "tma_info_core_ilp" }, @@ -718,20 +719,20 @@ "PublicDescription": "Fraction of Uops delivered by the DSB (aka D= ecoded ICache; or Uop Cache). Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_inst_mix_iptb, tma_lcp" }, { - "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= ", - "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "BriefDescription": "Average number of cycles of a switch from the= DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details= .", + "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWI= TCHES.PENALTY_CYCLES\\,cmask\\=3D1\\,edge@", "MetricGroup": "DSBmiss", "MetricName": "tma_info_frontend_dsb_switch_cost" }, { "BriefDescription": "Average number of Uops issued by front-end wh= en it issued something", - "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D0= x1@", + "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=3D1= @", "MetricGroup": "Fed;FetchBW", "MetricName": "tma_info_frontend_fetch_upc" }, { "BriefDescription": "Average Latency for L1 instruction cache miss= es", - "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D0x1\\,edge\\=3D0x1@", + "MetricExpr": "ICACHE_16B.IFDATA_STALL / cpu@ICACHE_16B.IFDATA_STA= LL\\,cmask\\=3D1\\,edge@", "MetricGroup": "Fed;FetchLat;IcMiss", "MetricName": "tma_info_frontend_icache_miss_latency" }, @@ -773,7 +774,7 @@ "MetricName": "tma_info_frontend_tbpc" }, { - "BriefDescription": "Branch instructions per taken branch", + "BriefDescription": "Branch instructions per taken branch.", "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR= _TAKEN", "MetricGroup": "Branches;Fed;PGO", "MetricName": "tma_info_inst_mix_bptkbranch" @@ -791,7 +792,7 @@ "MetricGroup": "Flops;InsType", "MetricName": "tma_info_inst_mix_iparith", "MetricThreshold": "tma_info_inst_mix_iparith < 10", - "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW" + "PublicDescription": "Instructions per FP Arithmetic instruction (= lower number means higher occurrence rate). Values < 1 are possible due to = intentional FMA double counting. Approximated prior to BDW." }, { "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bi= t instruction (lower number means higher occurrence rate)", @@ -799,7 +800,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx128", "MetricThreshold": "tma_info_inst_mix_iparith_avx128 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-b= it instruction (lower number means higher occurrence rate). Values < 1 are = possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit i= nstruction (lower number means higher occurrence rate)", @@ -807,7 +808,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx256", "MetricThreshold": "tma_info_inst_mix_iparith_avx256 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit = instruction (lower number means higher occurrence rate). Values < 1 are pos= sible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit in= struction (lower number means higher occurrence rate)", @@ -815,7 +816,7 @@ "MetricGroup": "Flops;FpVector;InsType", "MetricName": "tma_info_inst_mix_iparith_avx512", "MetricThreshold": "tma_info_inst_mix_iparith_avx512 < 10", - "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit i= nstruction (lower number means higher occurrence rate). Values < 1 are poss= ible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Double-= Precision instruction (lower number means higher occurrence rate)", @@ -823,7 +824,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_dp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_dp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Double= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per FP Arithmetic Scalar Single-= Precision instruction (lower number means higher occurrence rate)", @@ -831,7 +832,7 @@ "MetricGroup": "Flops;FpScalar;InsType", "MetricName": "tma_info_inst_mix_iparith_scalar_sp", "MetricThreshold": "tma_info_inst_mix_iparith_scalar_sp < 10", - "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting" + "PublicDescription": "Instructions per FP Arithmetic Scalar Single= -Precision instruction (lower number means higher occurrence rate). Values = < 1 are possible due to intentional FMA double counting." }, { "BriefDescription": "Instructions per Branch (lower number means h= igher occurrence rate)", @@ -886,7 +887,7 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO;tma_issueFB", "MetricName": "tma_info_inst_mix_iptb", - "MetricThreshold": "tma_info_inst_mix_iptb < 5 * 2 + 1", + "MetricThreshold": "tma_info_inst_mix_iptb < 11", "PublicDescription": "Instructions per taken branch. Related metri= cs: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidth= , tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_lcp" }, { @@ -1011,7 +1012,7 @@ }, { "BriefDescription": "Average Parallel L2 cache miss demand Loads", - "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D0x1@", + "MetricExpr": "OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD / cpu@O= FFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD\\,cmask\\=3D1@", "MetricGroup": "Memory_BW;Offcore", "MetricName": "tma_info_memory_latency_load_l2_mlp" }, @@ -1074,7 +1075,7 @@ }, { "BriefDescription": "", - "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D0x1@)", + "MetricExpr": "UOPS_EXECUTED.THREAD / (UOPS_EXECUTED.CORE_CYCLES_G= E_1 / 2 if #SMT_on else cpu@UOPS_EXECUTED.THREAD\\,cmask\\=3D1@)", "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", "MetricName": "tma_info_pipeline_execute" }, @@ -1101,12 +1102,12 @@ "MetricExpr": "INST_RETIRED.ANY / ASSISTS.ANY", "MetricGroup": "MicroSeq;Pipeline;Ret;Retire", "MetricName": "tma_info_pipeline_ipassist", - "MetricThreshold": "tma_info_pipeline_ipassist < 100000", + "MetricThreshold": "tma_info_pipeline_ipassist < 100e3", "PublicDescription": "Instructions per a microcode Assist invocati= on. See Assists tree node for details (lower number means higher occurrence= rate)" }, { - "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired", - "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D0x1@", + "BriefDescription": "Average number of Uops retired in cycles wher= e at least one uop has retired.", + "MetricExpr": "tma_retiring * tma_info_thread_slots / cpu@UOPS_RET= IRED.SLOTS\\,cmask\\=3D1@", "MetricGroup": "Pipeline;Ret", "MetricName": "tma_info_pipeline_retire" }, @@ -1147,14 +1148,13 @@ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", "MetricGroup": "Branches;OS", "MetricName": "tma_info_system_ipfarbranch", - "MetricThreshold": "tma_info_system_ipfarbranch < 1000000" + "MetricThreshold": "tma_info_system_ipfarbranch < 1e6" }, { "BriefDescription": "Cycles Per Instruction for the Operating Syst= em (OS) Kernel mode", "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", "MetricGroup": "OS", - "MetricName": "tma_info_system_kernel_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_system_kernel_cpi" }, { "BriefDescription": "Fraction of cycles spent in the Operating Sys= tem (OS) Kernel mode", @@ -1195,7 +1195,7 @@ "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / tma_info_core_core_= clks", "MetricGroup": "Power", "MetricName": "tma_info_system_power_license0_utilization", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for baseline license level 0. This includes non= -AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 1", @@ -1203,7 +1203,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license1_utilization", "MetricThreshold": "tma_info_system_power_license1_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 1. This includes high current= AVX 256-bit instructions as well as low current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of Core cycles where the core was ru= nning with power-delivery for license level 2 (introduced in SKX)", @@ -1211,7 +1211,7 @@ "MetricGroup": "Power", "MetricName": "tma_info_system_power_license2_utilization", "MetricThreshold": "tma_info_system_power_license2_utilization > 0= .5", - "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions" + "PublicDescription": "Fraction of Core cycles where the core was r= unning with power-delivery for license level 2 (introduced in SKX). This i= ncludes high current AVX 512-bit instructions." }, { "BriefDescription": "Fraction of cycles where both hardware Logica= l Processors were active", @@ -1239,7 +1239,7 @@ "MetricName": "tma_info_system_turbo_utilization" }, { - "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active", + "BriefDescription": "Per-Logical Processor actual clocks when the = Logical Processor is active.", "MetricExpr": "CPU_CLK_UNHALTED.THREAD", "MetricGroup": "Pipeline", "MetricName": "tma_info_thread_clks" @@ -1248,15 +1248,14 @@ "BriefDescription": "Cycles Per Instruction (per Logical Processor= )", "MetricExpr": "1 / tma_info_thread_ipc", "MetricGroup": "Mem;Pipeline", - "MetricName": "tma_info_thread_cpi", - "ScaleUnit": "1per_instr" + "MetricName": "tma_info_thread_cpi" }, { "BriefDescription": "The ratio of Executed- by Issued-Uops", "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", "MetricGroup": "Cor;Pipeline", "MetricName": "tma_info_thread_execute_per_issue", - "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage" + "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio= > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate o= f \"execute\" at rename stage." }, { "BriefDescription": "Instructions Per Cycle (per Logical Processor= )", @@ -1266,13 +1265,13 @@ }, { "BriefDescription": "Total issue-pipeline slots (per-Physical Core= till ICL; per-Logical Processor ICL onward)", - "MetricExpr": "slots", + "MetricExpr": "TOPDOWN.SLOTS", "MetricGroup": "TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots" }, { "BriefDescription": "Fraction of Physical Core issue-slots utilize= d by this Logical Processor", - "MetricExpr": "(tma_info_thread_slots / (slots / 2) if #SMT_on els= e 1)", + "MetricExpr": "(tma_info_thread_slots / (TOPDOWN.SLOTS / 2) if #SM= T_on else 1)", "MetricGroup": "SMT;TmaL1;tma_L1_group", "MetricName": "tma_info_thread_slots_utilization" }, @@ -1288,14 +1287,14 @@ "MetricExpr": "tma_retiring * tma_info_thread_slots / BR_INST_RETI= RED.NEAR_TAKEN", "MetricGroup": "Branches;Fed;FetchBW", "MetricName": "tma_info_thread_uptb", - "MetricThreshold": "tma_info_thread_uptb < 5 * 1.5" + "MetricThreshold": "tma_info_thread_uptb < 7.5" }, { - "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active", + "BriefDescription": "This metric represents fraction of cycles whe= re the Integer Divider unit was active.", "MetricExpr": "tma_divider - tma_fp_divider", "MetricGroup": "TopdownL4;tma_L4_group;tma_divider_group", "MetricName": "tma_int_divider", - "MetricThreshold": "tma_int_divider > 0.2 & tma_divider > 0.2 & tm= a_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_int_divider > 0.2 & (tma_divider > 0.2 & (= tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "ScaleUnit": "100%" }, { @@ -1303,8 +1302,8 @@ "MetricExpr": "ICACHE_TAG.STALLS / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;MemoryTLB;TopdownL3;tma= _L3_group;tma_fetch_latency_group", "MetricName": "tma_itlb_misses", - "MetricThreshold": "tma_itlb_misses > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS, FRONTEND_RETIRED.ITLB_MISS", + "MetricThreshold": "tma_itlb_misses > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Instruction TLB (ITLB) misses. Sample with: FRONTE= ND_RETIRED.STLB_MISS_PS;FRONTEND_RETIRED.ITLB_MISS_PS", "ScaleUnit": "100%" }, { @@ -1312,7 +1311,7 @@ "MetricExpr": "max((CYCLE_ACTIVITY.STALLS_MEM_ANY - CYCLE_ACTIVITY= .STALLS_L1D_MISS) / tma_info_thread_clks, 0)", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_issueL1;tma_issueMC;tma_memory_bound_group", "MetricName": "tma_l1_bound", - "MetricThreshold": "tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & = tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 &= tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled without loads missing the L1 Data (L1D) cache. The L1D cache typic= ally has the shortest latency. However; in certain cases like loads blocke= d on older stores; a load might suffer due to high latency even though it i= s being satisfied by the L1D. Another example is loads who miss in the TLB.= These cases are characterized by execution unit stalls; while some non-com= pleted demand load lives in the machine without having that demand load mis= sing the L1 cache. Sample with: MEM_LOAD_RETIRED.L1_HIT. Related metrics: t= ma_clears_resteers, tma_machine_clears, tma_microcode_sequencer, tma_ms_swi= tches, tma_ports_utilized_1", "ScaleUnit": "100%" }, @@ -1321,7 +1320,7 @@ "MetricExpr": "min(2 * (MEM_INST_RETIRED.ALL_LOADS - MEM_LOAD_RETI= RED.FB_HIT - MEM_LOAD_RETIRED.L1_MISS) * 20 / 100, max(CYCLE_ACTIVITY.CYCLE= S_MEM_ANY - CYCLE_ACTIVITY.CYCLES_L1D_MISS, 0)) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_l1_bound= _group", "MetricName": "tma_l1_latency_dependency", - "MetricThreshold": "tma_l1_latency_dependency > 0.1 & tma_l1_bound= > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l1_latency_dependency > 0.1 & (tma_l1_boun= d > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric([SKL+] roughly; [LNL]) estimates= fraction of cycles with demand load accesses that hit the L1D cache. The s= hort latency of the L1D cache may be exposed in pointer-chasing memory acce= ss patterns as an example. Sample with: MEM_LOAD_RETIRED.L1_HIT", "ScaleUnit": "100%" }, @@ -1331,7 +1330,7 @@ "MetricExpr": "MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_= HIT / MEM_LOAD_RETIRED.L1_MISS) / (MEM_LOAD_RETIRED.L2_HIT * (1 + MEM_LOAD_= RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS) + L1D_PEND_MISS.FB_FULL_PERIODS)= * ((CYCLE_ACTIVITY.STALLS_L1D_MISS - CYCLE_ACTIVITY.STALLS_L2_MISS) / tma_= info_thread_clks)", "MetricGroup": "BvML;CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_= L3_group;tma_memory_bound_group", "MetricName": "tma_l2_bound", - "MetricThreshold": "tma_l2_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", "PublicDescription": "This metric estimates how often the CPU was = stalled due to L2 cache accesses by loads. Avoiding cache misses (i.e. L1 = misses/L2 hits) can improve the latency and increase performance. Sample wi= th: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1340,7 +1339,7 @@ "MetricExpr": "5 * tma_info_system_core_frequency * MEM_LOAD_RETIR= ED.L2_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2) / = tma_info_thread_clks", "MetricGroup": "MemoryLat;TopdownL4;tma_L4_group;tma_l2_bound_grou= p", "MetricName": "tma_l2_hit_latency", - "MetricThreshold": "tma_l2_hit_latency > 0.05 & tma_l2_bound > 0.0= 5 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_l2_hit_latency > 0.05 & (tma_l2_bound > 0.= 05 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wi= th demand load accesses that hit the L2 cache under unloaded scenarios (pos= sibly L2 latency limited). Avoiding L1 cache misses (i.e. L1 misses/L2 hit= s) will improve the latency. Sample with: MEM_LOAD_RETIRED.L2_HIT", "ScaleUnit": "100%" }, @@ -1350,17 +1349,17 @@ "MetricExpr": "(CYCLE_ACTIVITY.STALLS_L2_MISS - CYCLE_ACTIVITY.STA= LLS_L3_MISS) / tma_info_thread_clks", "MetricGroup": "CacheHits;MemoryBound;TmaL3mem;TopdownL3;tma_L3_gr= oup;tma_memory_bound_group", "MetricName": "tma_l3_bound", - "MetricThreshold": "tma_l3_bound > 0.05 & tma_memory_bound > 0.2 &= tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT", + "MetricThreshold": "tma_l3_bound > 0.05 & (tma_memory_bound > 0.2 = & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often the CPU was = stalled due to loads accesses to L3 cache or contended with a sibling Core.= Avoiding cache misses (i.e. L2 misses/L3 hits) can improve the latency an= d increase performance. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles with= demand load accesses that hit the L3 cache under unloaded scenarios (possi= bly L3 latency limited)", - "MetricExpr": "(22.5 * tma_info_system_core_frequency - 5 * tma_in= fo_system_core_frequency) * (MEM_LOAD_RETIRED.L3_HIT * (1 + MEM_LOAD_RETIRE= D.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2)) / tma_info_thread_clks", + "MetricExpr": "17.5 * tma_info_system_core_frequency * (MEM_LOAD_R= ETIRED.L3_HIT * (1 + MEM_LOAD_RETIRED.FB_HIT / MEM_LOAD_RETIRED.L1_MISS / 2= )) / tma_info_thread_clks", "MetricGroup": "BvML;MemoryLat;TopdownL4;tma_L4_group;tma_issueLat= ;tma_l3_bound_group", "MetricName": "tma_l3_hit_latency", - "MetricThreshold": "tma_l3_hit_latency > 0.1 & tma_l3_bound > 0.05= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT. Related metrics: tma_b= ottleneck_cache_memory_latency, tma_branch_resteers, tma_mem_latency, tma_s= tore_latency", + "MetricThreshold": "tma_l3_hit_latency > 0.1 & (tma_l3_bound > 0.0= 5 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles wit= h demand load accesses that hit the L3 cache under unloaded scenarios (poss= ibly L3 latency limited). Avoiding private cache misses (i.e. L2 misses/L3= hits) will improve the latency; reduce contention with sibling physical co= res and increase performance. Note the value of this node may overlap with= its siblings. Sample with: MEM_LOAD_RETIRED.L3_HIT_PS. Related metrics: tm= a_bottleneck_cache_memory_latency, tma_mem_latency", "ScaleUnit": "100%" }, { @@ -1368,18 +1367,18 @@ "MetricExpr": "DECODE.LCP / tma_info_thread_clks", "MetricGroup": "FetchLat;TopdownL3;tma_L3_group;tma_fetch_latency_= group;tma_issueFB", "MetricName": "tma_lcp", - "MetricThreshold": "tma_lcp > 0.05 & tma_fetch_latency > 0.1 & tma= _frontend_bound > 0.15", - "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. Related metr= ics: tma_dsb_switches, tma_fetch_bandwidth, tma_info_botlnk_l2_dsb_bandwidt= h, tma_info_botlnk_l2_dsb_misses, tma_info_frontend_dsb_coverage, tma_info_= inst_mix_iptb", + "MetricThreshold": "tma_lcp > 0.05 & (tma_fetch_latency > 0.1 & tm= a_frontend_bound > 0.15)", + "PublicDescription": "This metric represents fraction of cycles CP= U was stalled due to Length Changing Prefixes (LCPs). Using proper compiler= flags or Intel Compiler by default will certainly avoid this. #Link: Optim= ization Guide about LCP BKMs. Related metrics: tma_dsb_switches, tma_fetch_= bandwidth, tma_info_botlnk_l2_dsb_bandwidth, tma_info_botlnk_l2_dsb_misses,= tma_info_frontend_dsb_coverage, tma_info_inst_mix_iptb", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations , instructions that require = no more than one uop (micro-operation)", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring light-weight operations -- instructions that require= no more than one uop (micro-operation)", "MetricExpr": "max(0, tma_retiring - tma_heavy_operations)", "MetricGroup": "Retire;TmaL2;TopdownL2;tma_L2_group;tma_retiring_g= roup", "MetricName": "tma_light_operations", "MetricThreshold": "tma_light_operations > 0.6", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations , instructions that require= no more than one uop (micro-operation). This correlates with total number = of instructions used by the program. A uops-per-instruction (see UopPI metr= ic) ratio of 1 or less should be expected for decently optimized code runni= ng on Intel Core/Xeon products. While this often indicates efficient X86 in= structions were executed; high value does not necessarily mean better perfo= rmance cannot be achieved. ([ICL+] Note this may undercount due to approxim= ation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIST= ", + "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring light-weight operations -- instructions that requir= e no more than one uop (micro-operation). This correlates with total number= of instructions used by the program. A uops-per-instruction (see UopPI met= ric) ratio of 1 or less should be expected for decently optimized code runn= ing on Intel Core/Xeon products. While this often indicates efficient X86 i= nstructions were executed; high value does not necessarily mean better perf= ormance cannot be achieved. ([ICL+] Note this may undercount due to approxi= mation using indirect events; [ADL+] .). Sample with: INST_RETIRED.PREC_DIS= T", "ScaleUnit": "100%" }, { @@ -1396,7 +1395,7 @@ "MetricExpr": "tma_dtlb_load - tma_load_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_hit", - "MetricThreshold": "tma_load_stlb_hit > 0.05 & tma_dtlb_load > 0.1= & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_hit > 0.05 & (tma_dtlb_load > 0.= 1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2= )))", "ScaleUnit": "100%" }, { @@ -1404,31 +1403,31 @@ "MetricExpr": "DTLB_LOAD_MISSES.WALK_ACTIVE / tma_info_thread_clks= ", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_load_gro= up", "MetricName": "tma_load_stlb_miss", - "MetricThreshold": "tma_load_stlb_miss > 0.05 & tma_dtlb_load > 0.= 1 & tma_l1_bound > 0.1 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss > 0.05 & (tma_dtlb_load > 0= .1 & (tma_l1_bound > 0.1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.= 2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_1G / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_1g", - "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_1g > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPL= ETED_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_2m", - "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_2m > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data load accesses.", "MetricExpr": "tma_load_stlb_miss * DTLB_LOAD_MISSES.WALK_COMPLETE= D_4K / (DTLB_LOAD_MISSES.WALK_COMPLETED_4K + DTLB_LOAD_MISSES.WALK_COMPLETE= D_2M_4M + DTLB_LOAD_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_load_stlb_mis= s_group", "MetricName": "tma_load_stlb_miss_4k", - "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & tma_load_stlb_m= iss > 0.05 & tma_dtlb_load > 0.1 & tma_l1_bound > 0.1 & tma_memory_bound > = 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_load_stlb_miss_4k > 0.05 & (tma_load_stlb_= miss > 0.05 & (tma_dtlb_load > 0.1 & (tma_l1_bound > 0.1 & (tma_memory_boun= d > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1437,7 +1436,7 @@ "MetricExpr": "(16 * max(0, MEM_INST_RETIRED.LOCK_LOADS - L2_RQSTS= .ALL_RFO) + MEM_INST_RETIRED.LOCK_LOADS / MEM_INST_RETIRED.ALL_STORES * (10= * L2_RQSTS.RFO_HIT + min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTSTAN= DING.CYCLES_WITH_DEMAND_RFO))) / tma_info_thread_clks", "MetricGroup": "LockCont;Offcore;TopdownL4;tma_L4_group;tma_issueR= FO;tma_l1_bound_group", "MetricName": "tma_lock_latency", - "MetricThreshold": "tma_lock_latency > 0.2 & tma_l1_bound > 0.1 & = tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_lock_latency > 0.2 & (tma_l1_bound > 0.1 &= (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU spent handling cache misses due to lock operations. Due to the microa= rchitecture handling of locks; they are classified as L1_Bound regardless o= f what memory source satisfied them. Sample with: MEM_INST_RETIRED.LOCK_LOA= DS. Related metrics: tma_store_latency", "ScaleUnit": "100%" }, @@ -1447,7 +1446,7 @@ "MetricGroup": "FetchBW;LSD;TopdownL3;tma_L3_group;tma_fetch_bandw= idth_group", "MetricName": "tma_lsd", "MetricThreshold": "tma_lsd > 0.15 & tma_fetch_bandwidth > 0.2", - "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure", + "PublicDescription": "This metric represents Core fraction of cycl= es in which CPU was likely limited due to LSD (Loop Stream Detector) unit. = LSD typically does well sustaining Uop supply. However; in some rare cases= ; optimal uop-delivery could not be reached for small loops whose size (in = terms of number of uops) does not suit well the LSD structure.", "ScaleUnit": "100%" }, { @@ -1457,15 +1456,15 @@ "MetricName": "tma_machine_clears", "MetricThreshold": "tma_machine_clears > 0.1 & tma_bad_speculation= > 0.15", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches", + "PublicDescription": "This metric represents fraction of slots the= CPU has wasted due to Machine Clears. These slots are either wasted by uo= ps fetched prior to the clear; or stalls the out-of-order portion of the ma= chine needs to recover its state after the clear. For example; this can hap= pen due to memory ordering Nukes (e.g. Memory Disambiguation) or Self-Modif= ying-Code (SMC) nukes. Sample with: MACHINE_CLEARS.COUNT. Related metrics: = tma_bottleneck_memory_synchronization, tma_clears_resteers, tma_contested_a= ccesses, tma_data_sharing, tma_false_sharing, tma_l1_bound, tma_microcode_s= equencer, tma_ms_switches, tma_remote_cache", "ScaleUnit": "100%" }, { "BriefDescription": "This metric estimates fraction of cycles wher= e the core's performance was likely hurt due to approaching bandwidth limit= s of external memory - DRAM ([SPR-HBM] and/or HBM)", - "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D0x4@) / tma_info_thread_clks", + "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, cpu@OFFCORE_REQUESTS_O= UTSTANDING.ALL_DATA_RD\\,cmask\\=3D4@) / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_d= ram_bound_group;tma_issueBW", "MetricName": "tma_mem_bandwidth", - "MetricThreshold": "tma_mem_bandwidth > 0.2 & tma_dram_bound > 0.1= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_bandwidth > 0.2 & (tma_dram_bound > 0.= 1 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the core's performance was likely hurt due to approaching bandwidth limi= ts of external memory - DRAM ([SPR-HBM] and/or HBM). The underlying heuris= tic assumes that a similar off-core traffic is generated by all IA cores. T= his metric does not aggregate non-data-read requests by this logical proces= sor; requests from other IA Logical Processors/Physical Cores/sockets; or o= ther non-IA devices like GPU; hence the maximum external memory bandwidth l= imits may or may not be approached when this metric is flagged (see Uncore = counters for that). Related metrics: tma_bottleneck_cache_memory_bandwidth,= tma_fb_full, tma_info_system_dram_bw_use, tma_sq_full", "ScaleUnit": "100%" }, @@ -1474,7 +1473,7 @@ "MetricExpr": "min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUESTS_OUTST= ANDING.CYCLES_WITH_DATA_RD) / tma_info_thread_clks - tma_mem_bandwidth", "MetricGroup": "BvML;MemoryLat;Offcore;TopdownL4;tma_L4_group;tma_= dram_bound_group;tma_issueLat", "MetricName": "tma_mem_latency", - "MetricThreshold": "tma_mem_latency > 0.1 & tma_dram_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_mem_latency > 0.1 & (tma_dram_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates fraction of cycles whe= re the performance was likely hurt due to latency from external memory - DR= AM ([SPR-HBM] and/or HBM). This metric does not aggregate requests from ot= her Logical Processors/Physical Cores/sockets (see Uncore counters for that= ). Related metrics: tma_bottleneck_cache_memory_latency, tma_l3_hit_latency= ", "ScaleUnit": "100%" }, @@ -1485,11 +1484,11 @@ "MetricName": "tma_memory_bound", "MetricThreshold": "tma_memory_bound > 0.2 & tma_backend_bound > 0= .2", "MetricgroupNoGroup": "TopdownL2", - "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= ", + "PublicDescription": "This metric represents fraction of slots the= Memory subsystem within the Backend was a bottleneck. Memory Bound estima= tes fraction of slots where pipeline is likely stalled due to demand load o= r store instructions. This accounts mainly for (1) non-completed in-flight = memory demand loads which coincides with execution units starvation; in add= ition to (2) cases where stores could impose backpressure on the pipeline w= hen many of them get buffered at the same time (less common out of the two)= .", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations , uops for memory load or store ac= cesses", + "BriefDescription": "This metric represents fraction of slots wher= e the CPU was retiring memory operations -- uops for memory load or store a= ccesses.", "MetricConstraint": "NO_GROUP_EVENTS", "MetricExpr": "tma_light_operations * MEM_INST_RETIRED.ANY / INST_= RETIRED.ANY", "MetricGroup": "Pipeline;TopdownL3;tma_L3_group;tma_light_operatio= ns_group", @@ -1511,7 +1510,7 @@ "MetricExpr": "BR_MISP_RETIRED.ALL_BRANCHES / (BR_MISP_RETIRED.ALL= _BRANCHES + MACHINE_CLEARS.COUNT) * INT_MISC.CLEAR_RESTEER_CYCLES / tma_inf= o_thread_clks", "MetricGroup": "BadSpec;BrMispredicts;BvMP;TopdownL4;tma_L4_group;= tma_branch_resteers_group;tma_issueBM", "MetricName": "tma_mispredicts_resteers", - "MetricThreshold": "tma_mispredicts_resteers > 0.05 & tma_branch_r= esteers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_mispredicts_resteers > 0.05 & (tma_branch_= resteers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to Branch Resteers as a result of Branch Mispredictio= n at execution stage. Sample with: INT_MISC.CLEAR_RESTEER_CYCLES. Related m= etrics: tma_bottleneck_mispredictions, tma_branch_mispredicts, tma_info_bad= _spec_branch_misprediction_cost", "ScaleUnit": "100%" }, @@ -1526,24 +1525,24 @@ }, { "BriefDescription": "This metric represents fraction of cycles whe= re (only) 4 uops were delivered by the MITE pipeline", - "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D0x4@ - cpu@IDQ.MITE_= UOPS\\,cmask\\=3D0x5@) / tma_info_thread_clks", + "MetricExpr": "(cpu@IDQ.MITE_UOPS\\,cmask\\=3D4@ - cpu@IDQ.MITE_UO= PS\\,cmask\\=3D5@) / tma_info_thread_clks", "MetricGroup": "DSBmiss;FetchBW;TopdownL4;tma_L4_group;tma_mite_gr= oup", "MetricName": "tma_mite_4wide", - "MetricThreshold": "tma_mite_4wide > 0.05 & tma_mite > 0.1 & tma_f= etch_bandwidth > 0.2", + "MetricThreshold": "tma_mite_4wide > 0.05 & (tma_mite > 0.1 & tma_= fetch_bandwidth > 0.2)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued , the Count Do= main; [ADL+] cycles)", + "BriefDescription": "This metric estimates penalty in terms of per= centage of([SKL+] injected blend uops out of all Uops Issued -- the Count D= omain; [ADL+] cycles)", "MetricExpr": "UOPS_ISSUED.VECTOR_WIDTH_MISMATCH / UOPS_ISSUED.ANY= ", "MetricGroup": "TopdownL5;tma_L5_group;tma_issueMV;tma_ports_utili= zed_0_group", "MetricName": "tma_mixing_vectors", "MetricThreshold": "tma_mixing_vectors > 0.05", - "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued , the Count D= omain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investigat= ing. Read more in Appendix B1 of the Optimizations Guide for this topic. Re= lated metrics: tma_ms_switches", + "PublicDescription": "This metric estimates penalty in terms of pe= rcentage of([SKL+] injected blend uops out of all Uops Issued -- the Count = Domain; [ADL+] cycles). Usually a Mixing_Vectors over 5% is worth investiga= ting. Read more in Appendix B1 of the Optimizations Guide for this topic. R= elated metrics: tma_ms_switches", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details", - "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D0x1@ / tma_info_core_co= re_clks / 2", + "BriefDescription": "This metric represents Core fraction of cycle= s in which CPU was likely limited due to the Microcode Sequencer (MS) unit = - see Microcode_Sequencer node for details.", + "MetricExpr": "cpu@IDQ.MS_UOPS\\,cmask\\=3D1@ / tma_info_core_core= _clks / 2", "MetricGroup": "MicroSeq;TopdownL3;tma_L3_group;tma_fetch_bandwidt= h_group", "MetricName": "tma_ms", "MetricThreshold": "tma_ms > 0.05 & tma_fetch_bandwidth > 0.2", @@ -1554,7 +1553,7 @@ "MetricExpr": "3 * IDQ.MS_SWITCHES / tma_info_thread_clks", "MetricGroup": "FetchLat;MicroSeq;TopdownL3;tma_L3_group;tma_fetch= _latency_group;tma_issueMC;tma_issueMS;tma_issueMV;tma_issueSO", "MetricName": "tma_ms_switches", - "MetricThreshold": "tma_ms_switches > 0.05 & tma_fetch_latency > 0= .1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_ms_switches > 0.05 & (tma_fetch_latency > = 0.1 & tma_frontend_bound > 0.15)", "PublicDescription": "This metric estimates the fraction of cycles= when the CPU was stalled due to switches of uop delivery to the Microcode = Sequencer (MS). Commonly used instructions are optimized for delivery by th= e DSB (decoded i-cache) or MITE (legacy instruction decode) pipelines. Cert= ain operations cannot be handled natively by the execution pipeline; and mu= st be performed by microcode (small programs injected into the execution st= ream). Switching to the MS too often can negatively impact performance. The= MS is designated to deliver long uop flows required by CISC instructions l= ike CPUID; or uncommon conditions like Floating Point Assists when dealing = with Denormals. Sample with: IDQ.MS_SWITCHES. Related metrics: tma_bottlene= ck_irregular_overhead, tma_clears_resteers, tma_l1_bound, tma_machine_clear= s, tma_microcode_sequencer, tma_mixing_vectors, tma_serializing_operation", "ScaleUnit": "100%" }, @@ -1563,7 +1562,7 @@ "MetricExpr": "tma_light_operations * INST_RETIRED.NOP / (tma_reti= ring * tma_info_thread_slots)", "MetricGroup": "BvBO;Pipeline;TopdownL4;tma_L4_group;tma_other_lig= ht_ops_group", "MetricName": "tma_nop_instructions", - "MetricThreshold": "tma_nop_instructions > 0.1 & tma_other_light_o= ps > 0.3 & tma_light_operations > 0.6", + "MetricThreshold": "tma_nop_instructions > 0.1 & (tma_other_light_= ops > 0.3 & tma_light_operations > 0.6)", "PublicDescription": "This metric represents fraction of slots whe= re the CPU was retiring NOP (no op) instructions. Compilers often use NOPs = for certain address alignments - e.g. start address of a function or loop b= ody. Sample with: INST_RETIRED.NOP", "ScaleUnit": "100%" }, @@ -1578,19 +1577,19 @@ "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types)", + "BriefDescription": "This metric estimates fraction of slots the C= PU was stalled due to other cases of misprediction (non-retired x86 branche= s or other types).", "MetricExpr": "max(tma_branch_mispredicts * (1 - BR_MISP_RETIRED.A= LL_BRANCHES / (INT_MISC.CLEARS_COUNT - MACHINE_CLEARS.COUNT)), 0.0001)", "MetricGroup": "BrMispredicts;BvIO;TopdownL3;tma_L3_group;tma_bran= ch_mispredicts_group", "MetricName": "tma_other_mispredicts", - "MetricThreshold": "tma_other_mispredicts > 0.05 & tma_branch_misp= redicts > 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_mispredicts > 0.05 & (tma_branch_mis= predicts > 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= ", + "BriefDescription": "This metric represents fraction of slots the = CPU has wasted due to Nukes (Machine Clears) not related to memory ordering= .", "MetricExpr": "max(tma_machine_clears * (1 - MACHINE_CLEARS.MEMORY= _ORDERING / MACHINE_CLEARS.COUNT), 0.0001)", "MetricGroup": "BvIO;Machine_Clears;TopdownL3;tma_L3_group;tma_mac= hine_clears_group", "MetricName": "tma_other_nukes", - "MetricThreshold": "tma_other_nukes > 0.05 & tma_machine_clears > = 0.1 & tma_bad_speculation > 0.15", + "MetricThreshold": "tma_other_nukes > 0.05 & (tma_machine_clears >= 0.1 & tma_bad_speculation > 0.15)", "ScaleUnit": "100%" }, { @@ -1634,8 +1633,8 @@ "MetricExpr": "((tma_ports_utilized_0 * tma_info_thread_clks + (EX= E_ACTIVITY.1_PORTS_UTIL + tma_retiring * EXE_ACTIVITY.2_PORTS_UTIL)) / tma_= info_thread_clks if ARITH.DIVIDER_ACTIVE < CYCLE_ACTIVITY.STALLS_TOTAL - CY= CLE_ACTIVITY.STALLS_MEM_ANY else (EXE_ACTIVITY.1_PORTS_UTIL + tma_retiring = * EXE_ACTIVITY.2_PORTS_UTIL) / tma_info_thread_clks)", "MetricGroup": "PortsUtil;TopdownL3;tma_L3_group;tma_core_bound_gr= oup", "MetricName": "tma_ports_utilization", - "MetricThreshold": "tma_ports_utilization > 0.15 & tma_core_bound = > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations", + "MetricThreshold": "tma_ports_utilization > 0.15 & (tma_core_bound= > 0.1 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates fraction of cycles the= CPU performance was potentially limited due to Core computation issues (no= n divider-related). Two distinct categories can be attributed into this me= tric: (1) heavy data-dependency among contiguous instructions would manifes= t in this metric - such cases are often referred to as low Instruction Leve= l Parallelism (ILP). (2) Contention on some hardware execution unit other t= han Divider. For example; when there are too many multiply operations.", "ScaleUnit": "100%" }, { @@ -1643,8 +1642,8 @@ "MetricExpr": "EXE_ACTIVITY.EXE_BOUND_0_PORTS / tma_info_thread_cl= ks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_ports_utiliza= tion_group", "MetricName": "tma_ports_utilized_0", - "MetricThreshold": "tma_ports_utilized_0 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric", + "MetricThreshold": "tma_ports_utilized_0 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents fraction of cycles CP= U executed no uops on any execution port (Logical Processor cycles since IC= L, Physical Core cycles otherwise). Long-latency instructions like divides = may contribute to this metric.", "ScaleUnit": "100%" }, { @@ -1652,7 +1651,7 @@ "MetricExpr": "EXE_ACTIVITY.1_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issueL1;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_1", - "MetricThreshold": "tma_ports_utilized_1 > 0.2 & tma_ports_utiliza= tion > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_1 > 0.2 & (tma_ports_utiliz= ation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles wh= ere the CPU executed total of 1 uop per cycle on all execution ports (Logic= al Processor cycles since ICL, Physical Core cycles otherwise). This can be= due to heavy data-dependency among software instructions; or over oversubs= cribing a particular hardware resource. In some other cases with high 1_Por= t_Utilized and L1_Bound; this metric can point to L1 data-cache latency bot= tleneck that may not necessarily manifest with complete execution starvatio= n (due to the short L1 latency e.g. walking a linked list) - looking at the= assembly can be helpful. Sample with: EXE_ACTIVITY.1_PORTS_UTIL. Related m= etrics: tma_l1_bound", "ScaleUnit": "100%" }, @@ -1661,7 +1660,7 @@ "MetricExpr": "EXE_ACTIVITY.2_PORTS_UTIL / tma_info_thread_clks", "MetricGroup": "PortsUtil;TopdownL4;tma_L4_group;tma_issue2P;tma_p= orts_utilization_group", "MetricName": "tma_ports_utilized_2", - "MetricThreshold": "tma_ports_utilized_2 > 0.15 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_2 > 0.15 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 2 uops per cycle on all execution ports (Logical Proces= sor cycles since ICL, Physical Core cycles otherwise). Loop Vectorization = -most compilers feature auto-Vectorization options today- reduces pressure = on the execution ports as multiple elements are calculated with same uop. S= ample with: EXE_ACTIVITY.2_PORTS_UTIL. Related metrics: tma_fp_scalar, tma_= fp_vector, tma_fp_vector_128b, tma_fp_vector_256b, tma_fp_vector_512b, tma_= port_0, tma_port_1, tma_port_5, tma_port_6", "ScaleUnit": "100%" }, @@ -1670,14 +1669,14 @@ "MetricExpr": "UOPS_EXECUTED.CYCLES_GE_3 / tma_info_thread_clks", "MetricGroup": "BvCB;PortsUtil;TopdownL4;tma_L4_group;tma_ports_ut= ilization_group", "MetricName": "tma_ports_utilized_3m", - "MetricThreshold": "tma_ports_utilized_3m > 0.4 & tma_ports_utiliz= ation > 0.15 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_ports_utilized_3m > 0.4 & (tma_ports_utili= zation > 0.15 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles CP= U executed total of 3 or more uops per cycle on all execution ports (Logica= l Processor cycles since ICL, Physical Core cycles otherwise). Sample with:= UOPS_EXECUTED.CYCLES_GE_3", "ScaleUnit": "100%" }, { "BriefDescription": "This category represents fraction of slots ut= ilized by useful work i.e. issued uops that eventually get retired", "DefaultMetricgroupName": "TopdownL1", - "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * slots", + "MetricExpr": "topdown\\-retiring / (topdown\\-fe\\-bound + topdow= n\\-bad\\-spec + topdown\\-retiring + topdown\\-be\\-bound) + 0 * tma_info_= thread_slots", "MetricGroup": "BvUW;Default;TmaL1;TopdownL1;tma_L1_group", "MetricName": "tma_retiring", "MetricThreshold": "tma_retiring > 0.7 | tma_heavy_operations > 0.= 1", @@ -1690,7 +1689,7 @@ "MetricExpr": "RESOURCE_STALLS.SCOREBOARD / tma_info_thread_clks", "MetricGroup": "BvIO;PortsUtil;TopdownL3;tma_L3_group;tma_core_bou= nd_group;tma_issueSO", "MetricName": "tma_serializing_operation", - "MetricThreshold": "tma_serializing_operation > 0.1 & tma_core_bou= nd > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_serializing_operation > 0.1 & (tma_core_bo= und > 0.1 & tma_backend_bound > 0.2)", "PublicDescription": "This metric represents fraction of cycles th= e CPU issue-pipeline was stalled due to serializing operations. Instruction= s like CPUID; WRMSR or LFENCE serialize the out-of-order execution which ma= y limit performance. Sample with: RESOURCE_STALLS.SCOREBOARD. Related metri= cs: tma_ms_switches", "ScaleUnit": "100%" }, @@ -1699,7 +1698,7 @@ "MetricExpr": "140 * MISC_RETIRED.PAUSE_INST / tma_info_thread_clk= s", "MetricGroup": "TopdownL4;tma_L4_group;tma_serializing_operation_g= roup", "MetricName": "tma_slow_pause", - "MetricThreshold": "tma_slow_pause > 0.05 & tma_serializing_operat= ion > 0.1 & tma_core_bound > 0.1 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_slow_pause > 0.05 & (tma_serializing_opera= tion > 0.1 & (tma_core_bound > 0.1 & tma_backend_bound > 0.2))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to PAUSE Instructions. Sample with: MISC_RETIRED.PAUS= E_INST", "ScaleUnit": "100%" }, @@ -1709,7 +1708,7 @@ "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_split_loads", "MetricThreshold": "tma_split_loads > 0.3", - "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS", + "PublicDescription": "This metric estimates fraction of cycles han= dling memory load split accesses - load that cross 64-byte cache line bound= ary. Sample with: MEM_INST_RETIRED.SPLIT_LOADS_PS", "ScaleUnit": "100%" }, { @@ -1718,8 +1717,8 @@ "MetricExpr": "MEM_INST_RETIRED.SPLIT_STORES / tma_info_core_core_= clks", "MetricGroup": "TopdownL4;tma_L4_group;tma_issueSpSt;tma_store_bou= nd_group", "MetricName": "tma_split_stores", - "MetricThreshold": "tma_split_stores > 0.2 & tma_store_bound > 0.2= & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES", + "MetricThreshold": "tma_split_stores > 0.2 & (tma_store_bound > 0.= 2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric represents rate of split store a= ccesses. Consider aligning your data to the 64-byte cache line granularity= . Sample with: MEM_INST_RETIRED.SPLIT_STORES_PS. Related metrics: tma_port_= 4", "ScaleUnit": "100%" }, { @@ -1727,7 +1726,7 @@ "MetricExpr": "L1D_PEND_MISS.L2_STALL / tma_info_thread_clks", "MetricGroup": "BvMB;MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_i= ssueBW;tma_l3_bound_group", "MetricName": "tma_sq_full", - "MetricThreshold": "tma_sq_full > 0.3 & tma_l3_bound > 0.05 & tma_= memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_sq_full > 0.3 & (tma_l3_bound > 0.05 & (tm= a_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric measures fraction of cycles wher= e the Super Queue (SQ) was full taking into account all request-types and b= oth hardware SMT threads (Logical Processors). Related metrics: tma_bottlen= eck_cache_memory_bandwidth, tma_fb_full, tma_info_system_dram_bw_use, tma_m= em_bandwidth", "ScaleUnit": "100%" }, @@ -1736,8 +1735,8 @@ "MetricExpr": "EXE_ACTIVITY.BOUND_ON_STORES / tma_info_thread_clks= ", "MetricGroup": "MemoryBound;TmaL3mem;TopdownL3;tma_L3_group;tma_me= mory_bound_group", "MetricName": "tma_store_bound", - "MetricThreshold": "tma_store_bound > 0.2 & tma_memory_bound > 0.2= & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES", + "MetricThreshold": "tma_store_bound > 0.2 & (tma_memory_bound > 0.= 2 & tma_backend_bound > 0.2)", + "PublicDescription": "This metric estimates how often CPU was stal= led due to RFO store memory accesses; RFO store issue a read-for-ownership= request before the write. Even though store accesses do not typically stal= l out-of-order CPUs; there are few cases where stores can lead to actual st= alls. This metric will be flagged should RFO stores be a bottleneck. Sample= with: MEM_INST_RETIRED.ALL_STORES_PS", "ScaleUnit": "100%" }, { @@ -1746,8 +1745,8 @@ "MetricExpr": "13 * LD_BLOCKS.STORE_FORWARD / tma_info_thread_clks= ", "MetricGroup": "TopdownL4;tma_L4_group;tma_l1_bound_group", "MetricName": "tma_store_fwd_blk", - "MetricThreshold": "tma_store_fwd_blk > 0.1 & tma_l1_bound > 0.1 &= tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading", + "MetricThreshold": "tma_store_fwd_blk > 0.1 & (tma_l1_bound > 0.1 = & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric roughly estimates fraction of cy= cles when the memory subsystem had loads blocked since they could not forwa= rd data from earlier (in program order) overlapping stores. To streamline m= emory operations in the pipeline; a load can avoid waiting for memory if a = prior in-flight store is writing the data that the load wants to read (stor= e forwarding process). However; in some cases the load may be blocked for a= significant time pending the store forward. For example; when the prior st= ore is writing a smaller region than the load is reading.", "ScaleUnit": "100%" }, { @@ -1755,8 +1754,8 @@ "MetricExpr": "(L2_RQSTS.RFO_HIT * 10 * (1 - MEM_INST_RETIRED.LOCK= _LOADS / MEM_INST_RETIRED.ALL_STORES) + (1 - MEM_INST_RETIRED.LOCK_LOADS / = MEM_INST_RETIRED.ALL_STORES) * min(CPU_CLK_UNHALTED.THREAD, OFFCORE_REQUEST= S_OUTSTANDING.CYCLES_WITH_DEMAND_RFO)) / tma_info_thread_clks", "MetricGroup": "BvML;LockCont;MemoryLat;Offcore;TopdownL4;tma_L4_g= roup;tma_issueRFO;tma_issueSL;tma_store_bound_group", "MetricName": "tma_store_latency", - "MetricThreshold": "tma_store_latency > 0.1 & tma_store_bound > 0.= 2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", - "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_branch_resteers, tma_fb_full, tma_l= 3_hit_latency, tma_lock_latency", + "MetricThreshold": "tma_store_latency > 0.1 & (tma_store_bound > 0= .2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", + "PublicDescription": "This metric estimates fraction of cycles the= CPU spent handling L1D store misses. Store accesses usually less impact ou= t-of-order core performance; however; holding resources for longer time can= lead into undesired implications (e.g. contention on L1D fill-buffer entri= es - see FB_Full). Related metrics: tma_fb_full, tma_lock_latency", "ScaleUnit": "100%" }, { @@ -1773,7 +1772,7 @@ "MetricExpr": "tma_dtlb_store - tma_store_stlb_miss", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_hit", - "MetricThreshold": "tma_store_stlb_hit > 0.05 & tma_dtlb_store > 0= .05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > = 0.2", + "MetricThreshold": "tma_store_stlb_hit > 0.05 & (tma_dtlb_store > = 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound= > 0.2)))", "ScaleUnit": "100%" }, { @@ -1781,31 +1780,31 @@ "MetricExpr": "DTLB_STORE_MISSES.WALK_ACTIVE / tma_info_core_core_= clks", "MetricGroup": "MemoryTLB;TopdownL5;tma_L5_group;tma_dtlb_store_gr= oup", "MetricName": "tma_store_stlb_miss", - "MetricThreshold": "tma_store_stlb_miss > 0.05 & tma_dtlb_store > = 0.05 & tma_store_bound > 0.2 & tma_memory_bound > 0.2 & tma_backend_bound >= 0.2", + "MetricThreshold": "tma_store_stlb_miss > 0.05 & (tma_dtlb_store >= 0.05 & (tma_store_bound > 0.2 & (tma_memory_bound > 0.2 & tma_backend_boun= d > 0.2)))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 1 GB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_1G / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_1g", - "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_1g > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 2 or 4 MB page= s for data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_2M_4M / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_C= OMPLETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_2m", - "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_2m > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { - "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses", + "BriefDescription": "This metric estimates the fraction of cycles = to walk the memory paging structures to cache translation of 4 KB pages for= data store accesses.", "MetricExpr": "tma_store_stlb_miss * DTLB_STORE_MISSES.WALK_COMPLE= TED_4K / (DTLB_STORE_MISSES.WALK_COMPLETED_4K + DTLB_STORE_MISSES.WALK_COMP= LETED_2M_4M + DTLB_STORE_MISSES.WALK_COMPLETED_1G)", "MetricGroup": "MemoryTLB;TopdownL6;tma_L6_group;tma_store_stlb_mi= ss_group", "MetricName": "tma_store_stlb_miss_4k", - "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & tma_store_stlb= _miss > 0.05 & tma_dtlb_store > 0.05 & tma_store_bound > 0.2 & tma_memory_b= ound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_store_stlb_miss_4k > 0.05 & (tma_store_stl= b_miss > 0.05 & (tma_dtlb_store > 0.05 & (tma_store_bound > 0.2 & (tma_memo= ry_bound > 0.2 & tma_backend_bound > 0.2))))", "ScaleUnit": "100%" }, { @@ -1813,7 +1812,7 @@ "MetricExpr": "9 * OCR.STREAMING_WR.ANY_RESPONSE / tma_info_thread= _clks", "MetricGroup": "MemoryBW;Offcore;TopdownL4;tma_L4_group;tma_issueS= mSt;tma_store_bound_group", "MetricName": "tma_streaming_stores", - "MetricThreshold": "tma_streaming_stores > 0.2 & tma_store_bound >= 0.2 & tma_memory_bound > 0.2 & tma_backend_bound > 0.2", + "MetricThreshold": "tma_streaming_stores > 0.2 & (tma_store_bound = > 0.2 & (tma_memory_bound > 0.2 & tma_backend_bound > 0.2))", "PublicDescription": "This metric estimates how often CPU was stal= led due to Streaming store memory accesses; Streaming store optimize out a= read request required by RFO stores. Even though store accesses do not typ= ically stall out-of-order CPUs; there are few cases where stores can lead t= o actual stalls. This metric will be flagged should Streaming stores be a b= ottleneck. Sample with: OCR.STREAMING_WR.ANY_RESPONSE. Related metrics: tma= _fb_full", "ScaleUnit": "100%" }, @@ -1822,7 +1821,7 @@ "MetricExpr": "10 * BACLEARS.ANY / tma_info_thread_clks", "MetricGroup": "BigFootprint;BvBC;FetchLat;TopdownL4;tma_L4_group;= tma_branch_resteers_group", "MetricName": "tma_unknown_branches", - "MetricThreshold": "tma_unknown_branches > 0.05 & tma_branch_reste= ers > 0.05 & tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15", + "MetricThreshold": "tma_unknown_branches > 0.05 & (tma_branch_rest= eers > 0.05 & (tma_fetch_latency > 0.1 & tma_frontend_bound > 0.15))", "PublicDescription": "This metric represents fraction of cycles th= e CPU was stalled due to new branch address clears. These are fetched branc= hes the Branch Prediction Unit was unable to recognize (e.g. first time the= branch is fetched or hitting BTB capacity limit) hence called Unknown Bran= ches. Sample with: BACLEARS.ANY", "ScaleUnit": "100%" }, @@ -1831,8 +1830,8 @@ "MetricExpr": "tma_retiring * UOPS_EXECUTED.X87 / UOPS_EXECUTED.TH= READ", "MetricGroup": "Compute;TopdownL4;tma_L4_group;tma_fp_arith_group", "MetricName": "tma_x87_use", - "MetricThreshold": "tma_x87_use > 0.1 & tma_fp_arith > 0.2 & tma_l= ight_operations > 0.6", - "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint", + "MetricThreshold": "tma_x87_use > 0.1 & (tma_fp_arith > 0.2 & tma_= light_operations > 0.6)", + "PublicDescription": "This metric serves as an approximation of le= gacy x87 usage. It accounts for instructions beyond X87 FP arithmetic opera= tions; hence may be used as a thermometer to avoid X87 high usage and prefe= rably upgrade to modern ISA. See Tip under Tuning Hint.", "ScaleUnit": "100%" }, { --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 8ABFC190068 for ; Wed, 26 Mar 2025 00:10:05 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947807; cv=none; b=BmjewhLIH2qmKiqDR3s4UxDkA/4xmxGDKfY9FWGEelh6kcnmR2IXRojcRVK94D2b75ALfqwAbVPXmgFODqgCH18no86Qke8ubLh5jxsHLQMh6Ldgb+UNjzibq97bZaIEAOgpVFInjoG0w6Zbrcy1WubiT4j3/hE0T7VujgeFouQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947807; c=relaxed/simple; bh=9QtWxO8l2GpcAYbRMHA4inZL25MuZzaCf9TT8yzkl7E=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=dvRAsYnTzvLtoePUYClZjTRc744SHZ3iSGnRiYohu84Ulu9FFOt3GckKfpPj/ImRqUK546FagEtb6gr6V0g2YsxMADCqf8/hXEgaUibH9BF4XN6//drorT34FOYCH3hjFM0zgYxQ/uFQQ0Q3CH4VWu5BArftmnyaDuoOkX3wizo= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=VlaCkWvD; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="VlaCkWvD" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-2fec3e38c2dso16425819a91.2 for ; Tue, 25 Mar 2025 17:10:05 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947805; x=1743552605; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=bjcCoWPLeFv7HToZ+VPWd4Z16z9IeZ5E4tfOnaC3rvg=; b=VlaCkWvDzEj6lrTpYUy8YCGJEEVYdqCU6eaxVTyQXMAD9tQweUUcyOp6ZkfoIHMCox xDS9cIyaQmd+jBJHqCbB2O32AdAPqbTRd2pXfe9dyhzQWabpu6KxiEhY27MUCMhs6z0H DbNIsgN+WaEGUks5JBC4yvX+/D/xrAuOdgfrpaD5bu3TpisferpLXxRwIykXAWwgxbim KUjpeTvORZdVmv3ZsA0AdSA/YmlJH/TR1v8tlm72K8MBaFpi+2JMADJNkm6Fc+f/hx6t ix8oQJrbhSGBRtKT0FYufgTOhc4GNskLnb+8Qkj7uU9Xi+UbOOxyqp6Tg2Yujdk/P0AP ByAQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947805; x=1743552605; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=bjcCoWPLeFv7HToZ+VPWd4Z16z9IeZ5E4tfOnaC3rvg=; b=qKFdZh1oB9kULe8BeIHSu1UZ1AZvVxnDz0LQ5cOQfHAfZyIr5XiqxCPqF7oBDuj8A0 po+dYw36Hk+918uxzfNKar535skBn6hoW7i3UDJtZEWW6kkuTImynie0gs9a5JsW4RMa BAfQjeXrtlz7xi/yNKXMmCa/IyftpmuLWsky0vpjM5sZ0/AFF+Aj0TFFT2as9r4OX/hW G9b5cniKV2ipr503F6H1l6VhzpobcXZ6YHLS1jj7QsEkTc0yYETHMsu4++MVwISSXRVx l9vPOVvmJO7lburK7W2wodBma07HKEkZw/4WVxXCM7xwh46dTZI8QhC3pO/KOkERwpzQ GnzQ== X-Forwarded-Encrypted: i=1; AJvYcCU5piI/bREQJwlUOae5gsC2BErKKmeAllXB7Qq2MW9n8nxtswaIBL1N4x6LEn0CbpRIEgJOhQLN9a+hU8E=@vger.kernel.org X-Gm-Message-State: AOJu0YybnPPMoO5imlY1U2y/sRDFXuM14YS+fZ/j260gHYY5WSGVu4ME fg+zUzjLdtq7yXDVB+LZn5okm5nNP4InsdIllnSR/SEGolOWc1xadHfQFj4M+uI8UMBqr8GaKbi x4qyLBw== X-Google-Smtp-Source: AGHT+IGPKt0S0YUd4fGnI98Ax+NYjlpCe5sPqtoR7BrjF4gq5QfezF8MF2KvT76zmSRO5elsl+UjMSzged09 X-Received: from pgac3.prod.google.com ([2002:a05:6a02:2943:b0:af9:775:2e8b]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a21:688:b0:1f5:58b9:6d9b with SMTP id adf61e73a8af0-1fe42f2c91dmr31426051637.12.1742947804799; Tue, 25 Mar 2025 17:10:04 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:31 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-34-irogers@google.com> Subject: [PATCH v3 33/35] perf vendor events: Update westmereep-dp events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and virtual memory. Signed-off-by: Ian Rogers --- .../arch/x86/westmereep-dp/cache.json | 32 +++++++++++++++ .../arch/x86/westmereep-dp/other.json | 40 ------------------- .../x86/westmereep-dp/virtual-memory.json | 8 ++++ 3 files changed, 40 insertions(+), 40 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/westmereep-dp/cache.json b/tool= s/perf/pmu-events/arch/x86/westmereep-dp/cache.json index 30845c7dbf08..f6f95f3ff301 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-dp/cache.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-dp/cache.json @@ -119,6 +119,38 @@ "SampleAfterValue": "100000", "UMask": "0x2" }, + { + "BriefDescription": "L1I instruction fetch stall cycles", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.CYCLES_STALLED", + "SampleAfterValue": "2000000", + "UMask": "0x4" + }, + { + "BriefDescription": "L1I instruction fetch hits", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.HITS", + "SampleAfterValue": "2000000", + "UMask": "0x1" + }, + { + "BriefDescription": "L1I instruction fetch misses", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.MISSES", + "SampleAfterValue": "2000000", + "UMask": "0x2" + }, + { + "BriefDescription": "L1I Instruction fetches", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.READS", + "SampleAfterValue": "2000000", + "UMask": "0x3" + }, { "BriefDescription": "All L2 data requests", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereep-dp/other.json b/tool= s/perf/pmu-events/arch/x86/westmereep-dp/other.json index bcf5bcf637c0..c0cf8bae8074 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-dp/other.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-dp/other.json @@ -15,46 +15,6 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, - { - "BriefDescription": "L1I instruction fetch stall cycles", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.CYCLES_STALLED", - "SampleAfterValue": "2000000", - "UMask": "0x4" - }, - { - "BriefDescription": "L1I instruction fetch hits", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.HITS", - "SampleAfterValue": "2000000", - "UMask": "0x1" - }, - { - "BriefDescription": "L1I instruction fetch misses", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.MISSES", - "SampleAfterValue": "2000000", - "UMask": "0x2" - }, - { - "BriefDescription": "L1I Instruction fetches", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.READS", - "SampleAfterValue": "2000000", - "UMask": "0x3" - }, - { - "BriefDescription": "Large ITLB hit", - "Counter": "0,1,2,3", - "EventCode": "0x82", - "EventName": "LARGE_ITLB.HIT", - "SampleAfterValue": "200000", - "UMask": "0x1" - }, { "BriefDescription": "Loads that partially overlap an earlier store= ", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereep-dp/virtual-memory.js= on b/tools/perf/pmu-events/arch/x86/westmereep-dp/virtual-memory.json index 53d7f76325a3..84c920637b12 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-dp/virtual-memory.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-dp/virtual-memory.json @@ -152,6 +152,14 @@ "SampleAfterValue": "200000", "UMask": "0x20" }, + { + "BriefDescription": "Large ITLB hit", + "Counter": "0,1,2,3", + "EventCode": "0x82", + "EventName": "LARGE_ITLB.HIT", + "SampleAfterValue": "200000", + "UMask": "0x1" + }, { "BriefDescription": "Retired loads that miss the DTLB (Precise Eve= nt)", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 3BEC8137742 for ; Wed, 26 Mar 2025 00:10:17 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947819; cv=none; b=iKZTp41QGxO2RMb6Lsty6MUySZqW2IbdmnZKeGHoYVEQJwx/14aODsLF2biOzX0fW87mylIejZgHdyJbLfPWlQ/FEu5vBHPR21Fp6z69FzngqTGCOaRqxRlQlLpPU6jX4Kl7seCQFh3TZQLlhj845D5iphOORfkQomB11f7y/AA= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947819; c=relaxed/simple; bh=9yS94I0V+aj5EaLixu70v/BvSVLiAUj4x7/o+VoVa7U=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=ex/ohtilZV3lWogW51kjxSncEVZY0BipJmmYe6BuuQEAgZL0XCKV4GLyoclp4SfKXWCJ3FtUANmUkWslLaf+unJPbp21BTWpgUFG5y89mh8VGRriLbvFREdp5QzeGS6Crt7sf4CsZpznVEq6Gr8IH8MupZFfDmSoJylPYl5a3n8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=RqoGBz+E; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="RqoGBz+E" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-2ff53a4754aso16909808a91.2 for ; Tue, 25 Mar 2025 17:10:17 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947817; x=1743552617; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=RPugHL8eCvwfstte7zHSdBSJ9ey/olpTRBhxpg/Uv8M=; b=RqoGBz+E7ckHDTH86lALFdudhtnuDiQKfyXLIdLxSQgUrVRd5EjDRQv2ABmhbm3KWZ 2byMdaZNgdELWQCmNoBBTdmGmKnQHHsKw+ztD6ffnGJPSPqnpDnvcKLgrk4JPt/c+ipd 5KpjcNb+c50w0l8YetKPEgcPU2+aIFeM3UTKXwGN2XRJcXLpVAs1Wv8rG77CA8DmHJ/n Ye3eP/6jp9ktbZObU+FqKxk9wvG+SDfuDUYOg7EAYTAnm1Q8/Eqc92ODN0bnCbQOdYHU xW3HtYw4BuTmyYIqHSG0hgMjtptvm5PudF4wSJ7M+LrlM7vYNHY39sMTNHZlhMtW7KFu NqHA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947817; x=1743552617; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=RPugHL8eCvwfstte7zHSdBSJ9ey/olpTRBhxpg/Uv8M=; b=M0sVcxfK7xwqJe7i/rELFG4y/OtO5RH5AOfhZwSclNe+iyIUEYZZLqcLEWgX9s2JHs BkwFIYj7WulBK0m6I4SvPEg3ztuSbUslmMtkjDAFy81nfcPwaEbHcR5pqUKtpIO9B0M3 MWPvp5kB0z167mWDlWbR1ZOqgu0lJDvxQDFIewh+QofleQbdR9hSrLk89VDWxepAI1df AmYMUbU8nzMU3HGOBSZSoWnygCC9lYSKuqlEfzQS83DBcPRWn6IOMszzyfjW+sF1Ee+x S+8VFERHSKDLLijhDGuLz0cg5FpNqJgxPD7WMVda2R4Ld4+FWIi88SyqPqB0EplRr4mG chHw== X-Forwarded-Encrypted: i=1; AJvYcCW/DztAFLWwTQ+mxXr5STG0/5y7TzZP/QYPFI3ovc4tasCLz638BhoeYXQEi9R1AW/ShoexPXKWsPfjNdg=@vger.kernel.org X-Gm-Message-State: AOJu0YxVXVUt5Hdau4NwtSkI0yLdS8WFp2gRMldkNzqQB4/keIfbYmr2 jTJFVQC13AlHr7tQs44ufOTpLla25MVkNskTRjducE25sI22nY2VYAjyhrKpaPfmBcha/F1y05F Z0HiTuA== X-Google-Smtp-Source: AGHT+IHB1Assd3GXiq7n7MR3LAdzwPPD57kW6fdNmjyVWfM8Ut/1geDGx/Lt8tX0rb7MoadYj0eZjOwOJvCr X-Received: from pgjg11.prod.google.com ([2002:a63:dd4b:0:b0:af8:eb05:4022]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6a20:6a0e:b0:1f5:9098:e448 with SMTP id adf61e73a8af0-1fe42f2a0bemr23636329637.5.1742947816712; Tue, 25 Mar 2025 17:10:16 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:32 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-35-irogers@google.com> Subject: [PATCH v3 34/35] perf vendor events: Update westmereep-dp events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and virtual memory. Signed-off-by: Ian Rogers --- .../arch/x86/westmereep-sp/cache.json | 32 +++++++++++++++ .../arch/x86/westmereep-sp/other.json | 40 ------------------- .../x86/westmereep-sp/virtual-memory.json | 8 ++++ 3 files changed, 40 insertions(+), 40 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/westmereep-sp/cache.json b/tool= s/perf/pmu-events/arch/x86/westmereep-sp/cache.json index 90cb367f5798..0cd571472dca 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-sp/cache.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-sp/cache.json @@ -119,6 +119,38 @@ "SampleAfterValue": "100000", "UMask": "0x2" }, + { + "BriefDescription": "L1I instruction fetch stall cycles", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.CYCLES_STALLED", + "SampleAfterValue": "2000000", + "UMask": "0x4" + }, + { + "BriefDescription": "L1I instruction fetch hits", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.HITS", + "SampleAfterValue": "2000000", + "UMask": "0x1" + }, + { + "BriefDescription": "L1I instruction fetch misses", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.MISSES", + "SampleAfterValue": "2000000", + "UMask": "0x2" + }, + { + "BriefDescription": "L1I Instruction fetches", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.READS", + "SampleAfterValue": "2000000", + "UMask": "0x3" + }, { "BriefDescription": "All L2 data requests", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereep-sp/other.json b/tool= s/perf/pmu-events/arch/x86/westmereep-sp/other.json index bcf5bcf637c0..c0cf8bae8074 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-sp/other.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-sp/other.json @@ -15,46 +15,6 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, - { - "BriefDescription": "L1I instruction fetch stall cycles", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.CYCLES_STALLED", - "SampleAfterValue": "2000000", - "UMask": "0x4" - }, - { - "BriefDescription": "L1I instruction fetch hits", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.HITS", - "SampleAfterValue": "2000000", - "UMask": "0x1" - }, - { - "BriefDescription": "L1I instruction fetch misses", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.MISSES", - "SampleAfterValue": "2000000", - "UMask": "0x2" - }, - { - "BriefDescription": "L1I Instruction fetches", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.READS", - "SampleAfterValue": "2000000", - "UMask": "0x3" - }, - { - "BriefDescription": "Large ITLB hit", - "Counter": "0,1,2,3", - "EventCode": "0x82", - "EventName": "LARGE_ITLB.HIT", - "SampleAfterValue": "200000", - "UMask": "0x1" - }, { "BriefDescription": "Loads that partially overlap an earlier store= ", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereep-sp/virtual-memory.js= on b/tools/perf/pmu-events/arch/x86/westmereep-sp/virtual-memory.json index e7affdf7f41b..a1b22c82a9bf 100644 --- a/tools/perf/pmu-events/arch/x86/westmereep-sp/virtual-memory.json +++ b/tools/perf/pmu-events/arch/x86/westmereep-sp/virtual-memory.json @@ -128,6 +128,14 @@ "SampleAfterValue": "200000", "UMask": "0x20" }, + { + "BriefDescription": "Large ITLB hit", + "Counter": "0,1,2,3", + "EventCode": "0x82", + "EventName": "LARGE_ITLB.HIT", + "SampleAfterValue": "200000", + "UMask": "0x1" + }, { "BriefDescription": "Retired loads that miss the DTLB (Precise Eve= nt)", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog From nobody Fri Dec 19 10:43:35 2025 Received: from mail-pj1-f73.google.com (mail-pj1-f73.google.com [209.85.216.73]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 405EF19D09C for ; Wed, 26 Mar 2025 00:10:26 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.73 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947827; cv=none; b=DHwYOhQpSPVQe5nMYqpYqMEqRz8cF2mPSFg3QkfZZhP4fTilErCO+2bWBI/aCmlAlt1efpofslDjNCvXzbDmu2dNmw0hj7xH4r25nb6f1vvz5wr0EaegpfR4tKMidsYl33+KZN+jnfB30fGs9m/9mdkhJFBucWDfTcSK+UIMiXU= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1742947827; c=relaxed/simple; bh=ldgIIS6qJKA4PqUti5NYcAmN4SsICYce46h1OQ5emEo=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Content-Type; b=FgPflTM7ybLYSKXR8PtIuU6bpauBbShhKnvhLDpEUvlR0uvZIYeo0r2EFd3KajAnVeddz3bDzG+5x8KX7JTvP/wJBs03TXquaGqRK2rKCl4ZvJgjqs9EnMwlLn7jxZZm1ODYwUAgNtIo3grv4dkT1Tu/2IiiSSbWJV+I/Z2Kc+I= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=G0llqRrk; arc=none smtp.client-ip=209.85.216.73 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--irogers.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="G0llqRrk" Received: by mail-pj1-f73.google.com with SMTP id 98e67ed59e1d1-3032f4eacd8so5812254a91.3 for ; Tue, 25 Mar 2025 17:10:26 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1742947825; x=1743552625; darn=vger.kernel.org; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :from:to:cc:subject:date:message-id:reply-to; bh=XET9HH8V4TfXAFNsd7WcnaOZ0oJ/0MNCJs/F3W4WSf4=; b=G0llqRrkEjceaPgc7p+IXGUFW5jg1xOl36LeV4GzMA0EOefln0xvIFbAzKor1mLsy8 UlDWsvsE4J9ZRAQ0ypTe/tve6FK0blZVIh8eyhSMapCcLR+DadmitAfj+d+fB9m+wLLs x6vNw4ayD2+eM7GfXm7xTkdOQvg9xaKTha2qyJQZuouBmXo+jSak/JW8MxOA/cdQ9UJ9 5es7ySPLKmb2bUb7mIsc606bOFAfScn/Rg0UKUzB20soP+Xc967yogwvX/oTJjDSPZ0H Fk9P+kasGh9UKkxBMM7+SLQi2cC8wzj45FPHp8I0DES2s2UWP8eVqu4BXaTJ5pNodOun d5QQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1742947825; x=1743552625; h=to:from:subject:message-id:references:mime-version:in-reply-to:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=XET9HH8V4TfXAFNsd7WcnaOZ0oJ/0MNCJs/F3W4WSf4=; b=X/PdbBl9ZO3kjl51+b4IsAKWIl3FRvbbOKYwfCW8Rn6isejl+rksSoDCJqaomYrvjV 8Nkw3M2loR2hzOYtfLTc2in8JfgoWPYtFNtCcKrswB5btKqGtN0o1sNEpG1F/TZNnJzD uDbKTwmKSQUb1xb5GIpoAiggWhZQ5NZZ5NXYK1QbajbIC7ImBzK7hy6hqoWpMm/CX5uh tz1XYg/A5NKA4s0VIu1l4SeB3iGxbO4oNFEu14RmJ7gTC8YgBjDQwYk4tSyDsLT6hok8 X7uc8bvxK9NigiY83Hu5I8RIoRLhqpmPHhDkW26ICIBS4pOf9i3G5GtA4d0096yjqiHR 79eQ== X-Forwarded-Encrypted: i=1; AJvYcCXDpf6eaUzendriEWDmIYX5umL7uki/ngLOcxcWf1OIi3ViXNFPH3U+nO91BuMwiR7PCOXQyDs/VfhDaUc=@vger.kernel.org X-Gm-Message-State: AOJu0YyFku/wkhbYyTkPEqXC0nElQIQngf2y53XYzFhaYXLPOX2paEAa NUigu+F2jweTbaplpVpU0/f76oXu8W7FMmdN6bvi7S5a1ontALkI3yZ2q/FIe7rIgATF53zlLtF 7y7goFg== X-Google-Smtp-Source: AGHT+IHFG8eaTHAVrj6Wv9IPR8ViyWjLI07OlR/vqWh01D836EQzijuXH0iKYW8evnBwhKcdgQRbtEFSW0HL X-Received: from pjbsc11.prod.google.com ([2002:a17:90b:510b:b0:2f9:dc36:b11]) (user=irogers job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:350d:b0:2ee:c2b5:97a0 with SMTP id 98e67ed59e1d1-3030fefe3cemr25488985a91.25.1742947825624; Tue, 25 Mar 2025 17:10:25 -0700 (PDT) Date: Tue, 25 Mar 2025 17:04:33 -0700 In-Reply-To: <20250326000434.83325-1-irogers@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250326000434.83325-1-irogers@google.com> X-Mailer: git-send-email 2.49.0.395.g12beb8f557-goog Message-ID: <20250326000434.83325-36-irogers@google.com> Subject: [PATCH v3 35/35] perf vendor events: Update westmereep-dp events From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , Kan Liang , "=?UTF-8?q?Andreas=20F=C3=A4rber?=" , Manivannan Sadhasivam , Maxime Coquelin , Alexandre Torgue , Caleb Biggers , Weilin Wang , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Perry Taylor , Thomas Falcon Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Update event topic moving other topic events to cache and virtual memory. Signed-off-by: Ian Rogers --- .../pmu-events/arch/x86/westmereex/cache.json | 32 +++++++++++++++ .../pmu-events/arch/x86/westmereex/other.json | 40 ------------------- .../arch/x86/westmereex/virtual-memory.json | 8 ++++ 3 files changed, 40 insertions(+), 40 deletions(-) diff --git a/tools/perf/pmu-events/arch/x86/westmereex/cache.json b/tools/p= erf/pmu-events/arch/x86/westmereex/cache.json index 9f922370ee8b..2a677d10f688 100644 --- a/tools/perf/pmu-events/arch/x86/westmereex/cache.json +++ b/tools/perf/pmu-events/arch/x86/westmereex/cache.json @@ -119,6 +119,38 @@ "SampleAfterValue": "100000", "UMask": "0x2" }, + { + "BriefDescription": "L1I instruction fetch stall cycles", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.CYCLES_STALLED", + "SampleAfterValue": "2000000", + "UMask": "0x4" + }, + { + "BriefDescription": "L1I instruction fetch hits", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.HITS", + "SampleAfterValue": "2000000", + "UMask": "0x1" + }, + { + "BriefDescription": "L1I instruction fetch misses", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.MISSES", + "SampleAfterValue": "2000000", + "UMask": "0x2" + }, + { + "BriefDescription": "L1I Instruction fetches", + "Counter": "0,1,2,3", + "EventCode": "0x80", + "EventName": "L1I.READS", + "SampleAfterValue": "2000000", + "UMask": "0x3" + }, { "BriefDescription": "All L2 data requests", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereex/other.json b/tools/p= erf/pmu-events/arch/x86/westmereex/other.json index bcf5bcf637c0..c0cf8bae8074 100644 --- a/tools/perf/pmu-events/arch/x86/westmereex/other.json +++ b/tools/perf/pmu-events/arch/x86/westmereex/other.json @@ -15,46 +15,6 @@ "SampleAfterValue": "2000000", "UMask": "0x1" }, - { - "BriefDescription": "L1I instruction fetch stall cycles", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.CYCLES_STALLED", - "SampleAfterValue": "2000000", - "UMask": "0x4" - }, - { - "BriefDescription": "L1I instruction fetch hits", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.HITS", - "SampleAfterValue": "2000000", - "UMask": "0x1" - }, - { - "BriefDescription": "L1I instruction fetch misses", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.MISSES", - "SampleAfterValue": "2000000", - "UMask": "0x2" - }, - { - "BriefDescription": "L1I Instruction fetches", - "Counter": "0,1,2,3", - "EventCode": "0x80", - "EventName": "L1I.READS", - "SampleAfterValue": "2000000", - "UMask": "0x3" - }, - { - "BriefDescription": "Large ITLB hit", - "Counter": "0,1,2,3", - "EventCode": "0x82", - "EventName": "LARGE_ITLB.HIT", - "SampleAfterValue": "200000", - "UMask": "0x1" - }, { "BriefDescription": "Loads that partially overlap an earlier store= ", "Counter": "0,1,2,3", diff --git a/tools/perf/pmu-events/arch/x86/westmereex/virtual-memory.json = b/tools/perf/pmu-events/arch/x86/westmereex/virtual-memory.json index 0c3501e6e5a3..1800c6ecbf80 100644 --- a/tools/perf/pmu-events/arch/x86/westmereex/virtual-memory.json +++ b/tools/perf/pmu-events/arch/x86/westmereex/virtual-memory.json @@ -152,6 +152,14 @@ "SampleAfterValue": "200000", "UMask": "0x20" }, + { + "BriefDescription": "Large ITLB hit", + "Counter": "0,1,2,3", + "EventCode": "0x82", + "EventName": "LARGE_ITLB.HIT", + "SampleAfterValue": "200000", + "UMask": "0x1" + }, { "BriefDescription": "Retired loads that miss the DTLB (Precise Eve= nt)", "Counter": "0,1,2,3", --=20 2.49.0.395.g12beb8f557-goog