| /linux/tools/perf/pmu-events/arch/x86/haswell/ |
| H A D | hsw-metrics.json | 82 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
91 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
96 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
100 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
107 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
111 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
117 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
121 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
128 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
132 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/broadwellde/ |
| H A D | bdwde-metrics.json | 82 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
91 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
96 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
100 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
107 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
111 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
117 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
121 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
128 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: TOPDOWN.BR_MISPREDICT_SLOTS. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
132 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/broadwell/ |
| H A D | bdw-metrics.json | 82 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
91 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
96 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
100 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
107 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
111 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
117 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
121 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
128 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
132 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/ivybridge/ |
| H A D | ivb-metrics.json | 82 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
91 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
96 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
100 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
107 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
111 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
117 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
121 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
128 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
132 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/amdzen4/ |
| H A D | pipeline.json | 9 "BriefDescription": "Fraction of dispatch slots that remained unused because the frontend did not supply enough instructions/ops.",
16 "BriefDescription": "Fraction of dispatched ops that did not retire.",
23 "BriefDescription": "Fraction of dispatch slots that remained unused because of backend stalls.",
30 "BriefDescription": "Fraction of dispatch slots that remained unused because the other thread was selected.",
37 "BriefDescription": "Fraction of dispatch slots used by ops that retired.",
44 "BriefDescription": "Fraction of dispatch slots that remained unused because of a latency bottleneck in the frontend (such as instruction cache or TLB misses).",
51 "BriefDescription": "Fraction of dispatch slots that remained unused because of a bandwidth bottleneck in the frontend (such as decode or op cache fetch bandwidth).",
58 "BriefDescription": "Fraction of dispatched ops that were flushed due to branch mispredicts.",
65 "BriefDescription": "Fraction of dispatched ops that were flushed due to pipeline restarts (resyncs).",
72 "BriefDescription": "Fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/rocketlake/ |
| H A D | rkl-metrics.json | 97 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
105 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
110 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
114 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
121 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
125 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
132 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
229 "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.",
237 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
243 "PublicDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/tigerlake/ |
| H A D | tgl-metrics.json | 97 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
105 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
110 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
114 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
121 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
125 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
132 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
229 "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.",
237 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
243 "PublicDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/icelake/ |
| H A D | icl-metrics.json | 97 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
105 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
110 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
114 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
121 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
125 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
132 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
229 "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.",
237 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
243 "PublicDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/sandybridge/ |
| H A D | snb-metrics.json | 73 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
80 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
84 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
90 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
94 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
101 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
105 "BriefDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers",
110 "PublicDescription": "This metric represents fraction of cycles the CPU was stalled due to Branch Resteers. Branch Resteers estimates the Frontend delay in fetching operations from corrected path; following all sorts of miss-predicted branches. For example; branchy code with lots of miss-predictions might get categorized under Branch Resteers. Note the value of this node may overlap with its siblings. Sample with: BR_MISP_RETIRED.ALL_BRANCHES",
114 "BriefDescription": "This metric represents fraction of slots where Core non-memory issues were of a bottleneck",
121 "PublicDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/skylake/ |
| H A D | skl-metrics.json | 82 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
90 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
95 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
99 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
105 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
109 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
115 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
212 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
219 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
223 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/broadwellx/ |
| H A D | bdx-metrics.json | 284 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
293 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
298 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
302 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
309 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
313 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
319 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
323 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
330 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
334 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/graniterapids/ |
| H A D | gnr-metrics.json | 349 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
357 "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations",
365 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
370 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
374 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
382 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
389 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
393 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
400 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
497 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/haswellx/ |
| H A D | hsx-metrics.json | 284 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
293 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
298 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
302 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
309 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
313 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
319 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
323 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
330 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
334 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/icelakex/ |
| H A D | icx-metrics.json | 343 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
351 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
356 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
360 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
367 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
371 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
378 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
475 "BriefDescription": "This metric represents fraction of slots where the CPU was retiring branch instructions.",
483 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
489 "PublicDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/cascadelakex/ |
| H A D | clx-metrics.json | 321 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
329 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
334 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
338 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
344 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
348 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
354 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
451 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
458 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
462 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/skylakex/ |
| H A D | skx-metrics.json | 303 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
311 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
316 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: OTHER_ASSISTS.ANY",
320 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
326 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound.",
330 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
336 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
433 "BriefDescription": "This metric represents fraction of slots the CPU has wasted due to Branch Misprediction",
440 "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 when the out-of-order part of the machine needs to recover its state from a speculative path. Sample with: BR_MISP_RETIRED.ALL_BRANCHES. Related metrics: tma_bottleneck_mispredictions, tma_info_bad_spec_branch_misprediction_cost, tma_mispredicts_resteers",
444 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/emeraldrapids/ |
| H A D | emr-metrics.json | 303 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
311 "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations",
319 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
324 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
328 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
336 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
343 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
347 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
354 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
451 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/sapphirerapids/ |
| H A D | spr-metrics.json | 363 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
371 "BriefDescription": "This metric estimates fraction of cycles where the Advanced Matrix eXtensions (AMX) execution engine was busy with tile (arithmetic) operations",
379 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
384 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
388 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
396 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
403 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
407 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
414 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
511 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/arm64/ampere/ampereone/ |
| H A D | metrics.json | 233 "BriefDescription": "Fraction of slots lost due to misspeculation",
241 "BriefDescription": "Fraction of slots retiring, useful work",
249 "BriefDescription": "Fraction of slots the CPU was stalled due to backend non-memory subsystem issues",
256 "BriefDescription": "Fraction of slots the CPU was stalled due to backend memory subsystem issues (cache/tlb miss)",
263 "BriefDescription": "Fraction of slots lost due to branch misprediciton",
270 "BriefDescription": "Fraction of slots the CPU did not dispatch at full bandwidth - able to dispatch partial slots only (1, 2, or 3 uops)",
277 "BriefDescription": "Fraction of slots the CPU was stalled due to frontend latency issues (cache/tlb miss); nothing to dispatch",
284 "BriefDescription": "Fraction of slots lost due to other/non-branch misprediction misspeculation",
291 "BriefDescription": "Fraction of execute slots utilized",
298 "BriefDescription": "Fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/meteorlake/ |
| H A D | mtl-metrics.json | 557 "BriefDescription": "Fraction of cycles spent in Kernel mode",
721 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
730 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
735 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
740 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
749 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
756 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
761 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
768 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
878 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/arrowlake/ |
| H A D | arl-metrics.json | 583 "BriefDescription": "Fraction of cycles spent in Kernel mode",
747 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
756 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
761 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
766 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
775 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
782 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
787 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
794 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
904 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/lunarlake/ |
| H A D | lnl-metrics.json | 571 "BriefDescription": "Fraction of cycles spent in Kernel mode",
728 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
737 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
742 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
747 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
756 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
763 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
768 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
775 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
885 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/arm64/ampere/ampereonex/ |
| H A D | metrics.json | 233 "BriefDescription": "Fraction of slots lost due to misspeculation",
241 "BriefDescription": "Fraction of slots retiring, useful work",
249 "BriefDescription": "Fraction of slots the CPU was stalled due to backend non-memory subsystem issues",
256 "BriefDescription": "Fraction of slots the CPU was stalled due to backend memory subsystem issues (cache/tlb miss)",
263 "BriefDescription": "Fraction of slots lost due to branch misprediciton",
270 "BriefDescription": "Fraction of slots the CPU did not dispatch at full bandwidth - able to dispatch partial slots only (1, 2, or 3 uops)",
277 "BriefDescription": "Fraction of slots the CPU was stalled due to frontend latency issues (cache/tlb miss); nothing to dispatch",
284 "BriefDescription": "Fraction of slots lost due to other/non-branch misprediction misspeculation",
291 "BriefDescription": "Fraction of execute slots utilized",
298 "BriefDescription": "Fraction o
[all...] |
| /linux/tools/perf/pmu-events/arch/x86/alderlake/ |
| H A D | adl-metrics.json | 560 "BriefDescription": "Fraction of cycles spent in Kernel mode",
724 "BriefDescription": "This metric represents Core fraction of cycles CPU dispatched uops on execution ports for ALU operations.",
733 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists",
738 "PublicDescription": "This metric estimates fraction of slots the CPU retired uops delivered by the Microcode_Sequencer as a result of Assists. Assists are long sequences of uops that are required in certain corner-cases for operations that cannot be handled natively by the execution pipeline. For example; when working with very small floating point values (so-called Denormals); the FP units are not set up to perform these operations natively. Instead; a sequence of instructions to perform the computation on the Denormals is injected into the pipeline. Since these microcode sequences might be dozens of uops long; Assists can be extremely deleterious to performance and they can be avoided in many cases. Sample with: ASSISTS.ANY",
743 "BriefDescription": "This metric estimates fraction of slots the CPU retired uops as a result of handing SSE to AVX* or AVX* to SSE transition Assists.",
752 "BriefDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend",
759 "PublicDescription": "This category represents fraction of slots where no uops are being delivered due to a lack of required resources for accepting new uops in the Backend. Backend is the portion of the processor core where the out-of-order scheduler dispatches ready uops into their respective execution units; and once completed these uops get retired according to program order. For example; stalls due to data-cache misses or stalls due to the divider unit being overloaded are both categorized under Backend Bound. Backend Bound is further divided into two main categories: Memory Bound and Core Bound. Sample with: TOPDOWN.BACKEND_BOUND_SLOTS",
764 "BriefDescription": "This category represents fraction of slots wasted due to incorrect speculations",
771 "PublicDescription": "This category represents fraction of slots wasted due to incorrect speculations. This include slots used to issue uops that do not eventually get retired and slots for which the issue-pipeline was blocked due to recovery from earlier incorrect speculation. For example; wasted work due to miss-predicted branches are categorized under Bad Speculation category. Incorrect data speculation followed by Memory Ordering Nukes is another example.",
881 "BriefDescription": "This metric represents fraction o
[all...] |
| /linux/arch/m68k/fpsp040/ |
| H A D | binstr.S | 16 | bit 63. The fraction is multiplied by 10 using a mul by 2
26 | Copy the fraction in d2:d3 to d4:d5.
28 | A3. Multiply the fraction in d2:d3 by 8 using bit-field
32 | A4. Multiply the fraction in d4:d5 by 2 using shifts. The msb
51 | d2: upper 32-bits of fraction for mul by 8
52 | d3: lower 32-bits of fraction for mul by 8
53 | d4: upper 32-bits of fraction for mul by 2
54 | d5: lower 32-bits of fraction for mul by 2
84 movel %d2,%d4 |copy the fraction before muls
|