perf vendor events: Add metrics for Icelake Server
PATCH v3 1/1 perf vendor events: Add metrics for Icelake Server
Add JSON metrics for Icelake Server to perf.
Based on TMA metrics 4.21 at 01.org.
https://download.01.org/perfmon/
---
v3:
- PMU cstate_core and cstate_pkg are supported for ICX since 5.14-rc1, add cstate metrics for Core C1/C6 and Package C2/C6.
v2:
- Fix perf test 10 error.
# ./perf test 10 10: PMU events : 10.1: PMU event table sanity : Ok 10.2: PMU event map aliases : Ok 10.3: Parsing of PMU event table metrics : Ok 10.4: Parsing of PMU event table metrics with fake PMUs : Ok
- Remove cstate metrics because the kernel has not supported cstate_core and cstate_core for Icelake server.
- Remove the topdown L1/L2 metrics.
.../arch/x86/icelakex/icx-metrics.json | 315 ++++++++++++++++++ 1 file changed, 315 insertions(+) create mode 100644 tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +315 -0
@@ -0,0 +1,315 @@ +[ + { + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Summary", + "MetricName": "IPC" + }, + { + "BriefDescription": "Uops Per Instruction", + "MetricExpr": "UOPS_RETIRED.SLOTS / INST_RETIRED.ANY", + "MetricGroup": "Pipeline;Retire", + "MetricName": "UPI" + }, + { + "BriefDescription": "Instruction per taken branch", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;FetchBW;PGO", + "MetricName": "IpTB" + }, + { + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", + "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", + "MetricGroup": "Pipeline", + "MetricName": "CPI" + }, + { + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Pipeline", + "MetricName": "CLKS" + }, + { + "BriefDescription": "Instructions Per Cycle (per physical core)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT;TmaL1", + "MetricName": "CoreIPC" + }, + { + "BriefDescription": "Floating Point Operations Per Cycle", + "MetricExpr": "( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "Flops", + "MetricName": "FLOPc" + }, + { + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is at least 1 uop executed)", + "MetricExpr": "UOPS_EXECUTED.THREAD / ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 )", + "MetricGroup": "Pipeline;PortsUtil", + "MetricName": "ILP" + }, + { + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear)", + "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", + "MetricGroup": "BrMispredicts", + "MetricName": "IpMispredict" + }, + { + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", + "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "CORE_CLKS" + }, + { + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", + "MetricGroup": "InsType", + "MetricName": "IpLoad" + }, + { + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", + "MetricGroup": "InsType", + "MetricName": "IpStore" + }, + { + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", + "MetricGroup": "Branches;InsType", + "MetricName": "IpBranch" + }, + { + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", + "MetricGroup": "Branches", + "MetricName": "IpCall" + }, + { + "BriefDescription": "Branch instructions per taken branch. ", + "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;PGO", + "MetricName": "BpTkBranch" + }, + { + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", + "MetricGroup": "Flops;FpArith;InsType", + "MetricName": "IpFLOP" + }, + { + "BriefDescription": "Total number of retired Instructions, Sample with: INST_RETIRED.PREC_DIST", + "MetricExpr": "INST_RETIRED.ANY", + "MetricGroup": "Summary;TmaL1", + "MetricName": "Instructions" + }, + { + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", + "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "LSD", + "MetricName": "LSD_Coverage" + }, + { + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", + "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "DSB;FetchBW", + "MetricName": "DSB_Coverage" + }, + { + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand loads (in core cycles)", + "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", + "MetricGroup": "MemoryBound;MemoryLat", + "MetricName": "Load_Miss_Real_Latency" + }, + { + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)", + "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", + "MetricGroup": "MemoryBound;MemoryBW", + "MetricName": "MLP" + }, + { + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", + "MetricConstraint": "NO_NMI_WATCHDOG", + "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )", + "MetricGroup": "MemoryTLB", + "MetricName": "Page_Walks_Utilization" + }, + { + "BriefDescription": "Average data fill bandwidth to the L1 data cache [GB / sec]", + "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L1D_Cache_Fill_BW" + }, + { + "BriefDescription": "Average data fill bandwidth to the L2 cache [GB / sec]", + "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L2_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L3_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW;Offcore", + "MetricName": "L3_Cache_Access_BW" + }, + { + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L1MPKI" + }, + { + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L2MPKI" + }, + { + "BriefDescription": "L2 cache misses per kilo instruction for all request types (including speculative)", + "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD ) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses;Offcore", + "MetricName": "L2MPKI_All" + }, + { + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L3MPKI" + }, + { + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", + "MetricExpr": "1000 * L2_LINES_OUT.SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_Silent_PKI" + }, + { + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", + "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_NonSilent_PKI" + }, + { + "BriefDescription": "Average CPU Utilization", + "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@", + "MetricGroup": "HPC;Summary", + "MetricName": "CPU_Utilization" + }, + { + "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ / 1000000000 / duration_time", + "MetricGroup": "Summary;Power", + "MetricName": "Average_Frequency" + }, + { + "BriefDescription": "Giga Floating Point Operations Per Second", + "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / 1000000000 ) / duration_time", + "MetricGroup": "Flops;HPC", + "MetricName": "GFLOPs" + }, + { + "BriefDescription": "Average Frequency Utilization relative nominal frequency", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC", + "MetricGroup": "Power", + "MetricName": "Turbo_Utilization" + }, + { + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", + "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "SMT_2T_Utilization" + }, + { + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "OS", + "MetricName": "Kernel_Utilization" + }, + { + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", + "MetricExpr": "( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) / 1000000000 ) / duration_time", + "MetricGroup": "HPC;MemoryBW;SoC", + "MetricName": "DRAM_BW_Use" + }, + { + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches", + "MetricExpr": "1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( cha_0@event\\=0x0@ / duration_time )", + "MetricGroup": "MemoryLat;SoC", + "MetricName": "MEM_Read_Latency" + }, + { + "BriefDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@event\\=0x36\\,umask\\=0xC817FE01\\,thresh\\=1@", + "MetricGroup": "MemoryBW;SoC", + "MetricName": "MEM_Parallel_Reads" + }, + { + "BriefDescription": "Average latency of data read request to external 3D X-Point memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches", + "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / cha_0@event\\=0x0@ )", + "MetricGroup": "MemoryLat;SoC;Server", + "MetricName": "MEM_PMM_Read_Latency" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for reads [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe3@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Read_BW" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe7@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000000 / duration_time", + "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", + "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM + UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / 1000000000 / duration_time", + "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Read_BW" + }, + { + "BriefDescription": "Socket actual clocks when any core is active on that socket", + "MetricExpr": "cha_0@event\\=0x0@", + "MetricGroup": "SoC", + "MetricName": "Socket_CLKS" + }, + { + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", + "MetricGroup": "Branches;OS", + "MetricName": "IpFarBranch" + }, + { + "BriefDescription": "C1 residency percent per core", + "MetricExpr": "(cstate_core@c1\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C1_Core_Residency" + }, + { + "BriefDescription": "C6 residency percent per core", + "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Core_Residency" + }, + { + "BriefDescription": "C2 residency percent per package", + "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C2_Pkg_Residency" + }, + { + "BriefDescription": "C6 residency percent per package", + "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Pkg_Residency" + }, +]
Signature · 1 line
2.17.1
Hi Arnaldo,
Can you kindly pick up this patch?
Thanks
Jin Yao
Quoted 352 lines · Jin Yao
Add JSON metrics for Icelake Server to perf.
Based on TMA metrics 4.21 at 01.org.
https://download.01.org/perfmon/Signed-off-by Jin Yao <[email protected]>---
v3:
- PMUcstate_coreandcstate_pkgare supported for ICX since 5.14-rc1, add cstate metrics for Core C1/C6 and Package C2/C6.v2:
- Fix perf test 10 error.# ./perf test 10 10: PMU events : 10.1: PMU event table sanity : Ok 10.2: PMU event map aliases : Ok 10.3: Parsing of PMU event table metrics : Ok 10.4: Parsing of PMU event table metrics with fake PMUs : Ok- Remove cstate metrics because the kernel has not supported
cstate_coreandcstate_corefor Icelake server.- Remove the topdown L1/L2 metrics.
.../arch/x86/icelakex/icx-metrics.json | 315 ++++++++++++++++++ 1 file changed, 315 insertions(+)create mode 100644
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.jsontools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +315
-0@@ -0,0 +1,315 @@ +[ + { + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Summary", + "MetricName": "IPC" + }, + { + "BriefDescription": "Uops Per Instruction", + "MetricExpr": "UOPS_RETIRED.SLOTS / INST_RETIRED.ANY", + "MetricGroup": "Pipeline;Retire", + "MetricName": "UPI" + }, + { + "BriefDescription": "Instruction per taken branch", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;FetchBW;PGO", + "MetricName": "IpTB" + }, + { + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", + "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", + "MetricGroup": "Pipeline", + "MetricName": "CPI" + }, + { + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Pipeline", + "MetricName": "CLKS" + }, + { + "BriefDescription": "Instructions Per Cycle (per physical core)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT;TmaL1", + "MetricName": "CoreIPC" + }, + { + "BriefDescription": "Floating Point Operations Per Cycle", + "MetricExpr": "( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "Flops", + "MetricName": "FLOPc" + }, + { + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is at least 1 uop executed)", + "MetricExpr": "UOPS_EXECUTED.THREAD / ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 )", + "MetricGroup": "Pipeline;PortsUtil", + "MetricName": "ILP" + }, + { + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear)", + "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", + "MetricGroup": "BrMispredicts", + "MetricName": "IpMispredict" + }, + { + "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", + "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "CORE_CLKS" + }, + { + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", + "MetricGroup": "InsType", + "MetricName": "IpLoad" + }, + { + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", + "MetricGroup": "InsType", + "MetricName": "IpStore" + }, + { + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", + "MetricGroup": "Branches;InsType", + "MetricName": "IpBranch" + }, + { + "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", + "MetricGroup": "Branches", + "MetricName": "IpCall" + }, + { + "BriefDescription": "Branch instructions per taken branch. ", + "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;PGO", + "MetricName": "BpTkBranch" + }, + { + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", + "MetricGroup": "Flops;FpArith;InsType", + "MetricName": "IpFLOP" + }, + { + "BriefDescription": "Total number of retired Instructions, Sample with: INST_RETIRED.PREC_DIST", + "MetricExpr": "INST_RETIRED.ANY", + "MetricGroup": "Summary;TmaL1", + "MetricName": "Instructions" + }, + { + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", + "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "LSD", + "MetricName": "LSD_Coverage" + }, + { + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", + "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "DSB;FetchBW", + "MetricName": "DSB_Coverage" + }, + { + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand loads (in core cycles)", + "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", + "MetricGroup": "MemoryBound;MemoryLat", + "MetricName": "Load_Miss_Real_Latency" + }, + { + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)", + "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", + "MetricGroup": "MemoryBound;MemoryBW", + "MetricName": "MLP" + }, + { + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", + "MetricConstraint": "NO_NMI_WATCHDOG", + "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )", + "MetricGroup": "MemoryTLB", + "MetricName": "Page_Walks_Utilization" + }, + { + "BriefDescription": "Average data fill bandwidth to the L1 data cache [GB / sec]", + "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L1D_Cache_Fill_BW" + }, + { + "BriefDescription": "Average data fill bandwidth to the L2 cache [GB / sec]", + "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L2_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L3_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW;Offcore", + "MetricName": "L3_Cache_Access_BW" + }, + { + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L1MPKI" + }, + { + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L2MPKI" + }, + { + "BriefDescription": "L2 cache misses per kilo instruction for all request types (including speculative)", + "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD ) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses;Offcore", + "MetricName": "L2MPKI_All" + }, + { + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L3MPKI" + }, + { + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", + "MetricExpr": "1000 * L2_LINES_OUT.SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_Silent_PKI" + }, + { + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", + "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_NonSilent_PKI" + }, + { + "BriefDescription": "Average CPU Utilization", + "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@", + "MetricGroup": "HPC;Summary", + "MetricName": "CPU_Utilization" + }, + { + "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ / 1000000000 / duration_time", + "MetricGroup": "Summary;Power", + "MetricName": "Average_Frequency" + }, + { + "BriefDescription": "Giga Floating Point Operations Per Second", + "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / 1000000000 ) / duration_time", + "MetricGroup": "Flops;HPC", + "MetricName": "GFLOPs" + }, + { + "BriefDescription": "Average Frequency Utilization relative nominal frequency", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC", + "MetricGroup": "Power", + "MetricName": "Turbo_Utilization" + }, + { + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", + "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "SMT_2T_Utilization" + }, + { + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "OS", + "MetricName": "Kernel_Utilization" + }, + { + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", + "MetricExpr": "( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) / 1000000000 ) / duration_time", + "MetricGroup": "HPC;MemoryBW;SoC", + "MetricName": "DRAM_BW_Use" + }, + { + "BriefDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches", + "MetricExpr": "1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( cha_0@event\\=0x0@ / duration_time )", + "MetricGroup": "MemoryLat;SoC", + "MetricName": "MEM_Read_Latency" + }, + { + "BriefDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches", + "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@event\\=0x36\\,umask\\=0xC817FE01\\,thresh\\=1@", + "MetricGroup": "MemoryBW;SoC", + "MetricName": "MEM_Parallel_Reads" + }, + { + "BriefDescription": "Average latency of data read request to external 3D X-Point memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches", + "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / cha_0@event\\=0x0@ )", + "MetricGroup": "MemoryLat;SoC;Server", + "MetricName": "MEM_PMM_Read_Latency" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for reads [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe3@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Read_BW" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe7@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000000 / duration_time", + "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", + "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM + UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / 1000000000 / duration_time", + "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Read_BW" + }, + { + "BriefDescription": "Socket actual clocks when any core is active on that socket", + "MetricExpr": "cha_0@event\\=0x0@", + "MetricGroup": "SoC", + "MetricName": "Socket_CLKS" + }, + { + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", + "MetricGroup": "Branches;OS", + "MetricName": "IpFarBranch" + }, + { + "BriefDescription": "C1 residency percent per core", + "MetricExpr": "(cstate_core@c1\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C1_Core_Residency" + }, + { + "BriefDescription": "C6 residency percent per core", + "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Core_Residency" + }, + { + "BriefDescription": "C2 residency percent per package", + "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C2_Pkg_Residency" + }, + { + "BriefDescription": "C6 residency percent per package", + "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Pkg_Residency" + }, +]
Hi Andi,
I guess Arnaldo needs your ACK before consider picking up this patch.
Can you give an ACK for this patch?
Thanks
Jin Yao
Quoted 390 lines · Yao
Hi Arnaldo,
Can you kindly pick up this patch?
Thanks
Jin YaoQuoted 382 lines · Jin Yao
Add JSON metrics for Icelake Server to perf.
Based on TMA metrics 4.21 at 01.org.
https://download.01.org/perfmon/Signed-off-by Jin Yao <[email protected]>---
v3:
- PMUcstate_coreandcstate_pkgare supported for ICX since 5.14-rc1, add cstate metrics for Core C1/C6 and Package C2/C6.v2:
- Fix perf test 10 error.# ./perf test 10 10: PMU events : 10.1: PMU event table sanity : Ok 10.2: PMU event map aliases : Ok 10.3: Parsing of PMU event table metrics : Ok 10.4: Parsing of PMU event table metrics with fake PMUs : Ok- Remove cstate metrics because the kernel has not supported
cstate_coreandcstate_corefor Icelake server.- Remove the topdown L1/L2 metrics.
.../arch/x86/icelakex/icx-metrics.json | 315 ++++++++++++++++++
1 file changed, 315 insertions(+)create mode 100644
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.jsontools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +27
-0@@ -0,0 +1,315 @@ +[ + { + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Summary", + "MetricName": "IPC" + }, + { + "BriefDescription": "Uops Per Instruction", + "MetricExpr": "UOPS_RETIRED.SLOTS / INST_RETIRED.ANY", + "MetricGroup": "Pipeline;Retire", + "MetricName": "UPI" + }, + { + "BriefDescription": "Instruction per taken branch", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;FetchBW;PGO", + "MetricName": "IpTB" + }, + { + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", + "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", + "MetricGroup": "Pipeline", + "MetricName": "CPI" + }, + { + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor isactive.",
+ "MetricExpr": "CPU_CLK_UNHALTED.THREAD",
+ "MetricGroup": "Pipeline",
+ "MetricName": "CLKS"
+ },
+ {
+ "BriefDescription": "Instructions Per Cycle (per physical core)",
+ "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED",
+ "MetricGroup": "SMT;TmaL1",
+ "MetricName": "CoreIPC"
+ },
+ {
+ "BriefDescription": "Floating Point Operations Per Cycle",
+ "MetricExpr": "( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +
FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "Flops",
+ "MetricName": "FLOPc"
+ },
+ {
+ "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when
there is at least 1 uop executed)",
+ "MetricExpr": "UOPS_EXECUTED.THREAD / ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 )",
+ "MetricGroup": "Pipeline;PortsUtil",
+ "MetricName": "ILP"
+ },
+ {
+ "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction
(JEClear)",
+ "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES",
+ "MetricGroup": "BrMispredicts",
+ "MetricName": "IpMispredict"
+ },
+ {
+ "BriefDescription": "Core actual clocks when any Logical Processor is active on the
Physical Core",
+ "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED",
+ "MetricGroup": "SMT",
+ "MetricName": "CORE_CLKS"
+ },
+ {
+ "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)",
+ "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS",
+ "MetricGroup": "InsType",
+ "MetricName": "IpLoad"
+ },
+ {
+ "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)",
+ "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES",
+ "MetricGroup": "InsType",
+ "MetricName": "IpStore"
+ },
+ {
+ "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)",
+ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES",
+ "MetricGroup": "Branches;InsType",
+ "MetricName": "IpBranch"
+ },
+ {
+ "BriefDescription": "Instructions per (near) call (lower number means higher occurrence
rate)",
+ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL",
+ "MetricGroup": "Branches",
+ "MetricName": "IpCall"
+ },
+ {
+ "BriefDescription": "Branch instructions per taken branch. ",
+ "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN",
+ "MetricGroup": "Branches;PGO",
+ "MetricName": "BpTkBranch"
+ },
+ {
+ "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means
higher occurrence rate)",
+ "MetricExpr": "INST_RETIRED.ANY / ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +
FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", + "MetricGroup": "Flops;FpArith;InsType",
+ "MetricName": "IpFLOP"
+ },
+ {
+ "BriefDescription": "Total number of retired Instructions, Sample with:
INST_RETIRED.PREC_DIST",
+ "MetricExpr": "INST_RETIRED.ANY",
+ "MetricGroup": "Summary;TmaL1",
+ "MetricName": "Instructions"
+ },
+ {
+ "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka
Loop Cache)",
+ "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)",
+ "MetricGroup": "LSD",
+ "MetricName": "LSD_Coverage"
+ },
+ {
+ "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop
Cache)",
+ "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)",
+ "MetricGroup": "DSB;FetchBW",
+ "MetricName": "DSB_Coverage"
+ },
+ {
+ "BriefDescription": "Actual Average Latency for L1 data-cache miss demand loads (in core
cycles)",
+ "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS +
MEM_LOAD_RETIRED.FB_HIT )",
+ "MetricGroup": "MemoryBound;MemoryLat",
+ "MetricName": "Load_Miss_Real_Latency"
+ },
+ {
+ "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when
there is at least one such miss. Per-Logical Processor)",
+ "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES",
+ "MetricGroup": "MemoryBound;MemoryBW",
+ "MetricName": "MLP"
+ },
+ {
+ "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses
triggered by instruction/Load/Store accesses",
+ "MetricConstraint": "NO_NMI_WATCHDOG",
+ "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING +
DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )",
+ "MetricGroup": "MemoryTLB",
+ "MetricName": "Page_Walks_Utilization"
+ },
+ {
+ "BriefDescription": "Average data fill bandwidth to the L1 data cache [GB / sec]",
+ "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 /duration_time",
+ "MetricGroup": "MemoryBW",
+ "MetricName": "L1D_Cache_Fill_BW"
+ },
+ {
+ "BriefDescription": "Average data fill bandwidth to the L2 cache [GB / sec]",
+ "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 /duration_time",
+ "MetricGroup": "MemoryBW",
+ "MetricName": "L2_Cache_Fill_BW"
+ },
+ {
+ "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]",
+ "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 /duration_time",
+ "MetricGroup": "MemoryBW",
+ "MetricName": "L3_Cache_Fill_BW"
+ },
+ {
+ "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]",
+ "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 /duration_time",
+ "MetricGroup": "MemoryBW;Offcore",
+ "MetricName": "L3_Cache_Access_BW"
+ },
+ {
+ "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads",
+ "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY",
+ "MetricGroup": "CacheMisses",
+ "MetricName": "L1MPKI"
+ },
+ {
+ "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads",
+ "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY",
+ "MetricGroup": "CacheMisses",
+ "MetricName": "L2MPKI"
+ },
+ {
+ "BriefDescription": "L2 cache misses per kilo instruction for all request types
(including speculative)",
+ "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD
) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / INST_RETIRED.ANY",
+ "MetricGroup": "CacheMisses;Offcore",
+ "MetricName": "L2MPKI_All"
+ },
+ {
+ "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads",
+ "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY",
+ "MetricGroup": "CacheMisses",
+ "MetricName": "L3MPKI"
+ },
+ {
+ "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction
where the evicted lines are dropped (no writeback to L3 or memory)",
+ "MetricExpr": "1000 * L2_LINES_OUT.SILENT / INST_RETIRED.ANY",
+ "MetricGroup": "L2Evicts;Server",
+ "MetricName": "L2_Evictions_Silent_PKI"
+ },
+ {
+ "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction",
+ "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / INST_RETIRED.ANY",
+ "MetricGroup": "L2Evicts;Server",
+ "MetricName": "L2_Evictions_NonSilent_PKI"
+ },
+ {
+ "BriefDescription": "Average CPU Utilization",
+ "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@",
+ "MetricGroup": "HPC;Summary",
+ "MetricName": "CPU_Utilization"
+ },
+ {
+ "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]",
+ "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ /
1000000000 /duration_time",
+ "MetricGroup": "Summary;Power",
+ "MetricName": "Average_Frequency"
+ },
+ {
+ "BriefDescription": "Giga Floating Point Operations Per Second",
+ "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +
FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / 1000000000 ) /duration_time", + "MetricGroup": "Flops;HPC",
+ "MetricName": "GFLOPs"
+ },
+ {
+ "BriefDescription": "Average Frequency Utilization relative nominal frequency",
+ "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC",
+ "MetricGroup": "Power",
+ "MetricName": "Turbo_Utilization"
+ },
+ {
+ "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active",
+ "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED",
+ "MetricGroup": "SMT",
+ "MetricName": "SMT_2T_Utilization"
+ },
+ {
+ "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode",
+ "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD",
+ "MetricGroup": "OS",
+ "MetricName": "Kernel_Utilization"
+ },
+ {
+ "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]",
+ "MetricExpr": "( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) /
1000000000 ) /duration_time",
+ "MetricGroup": "HPC;MemoryBW;SoC",
+ "MetricName": "DRAM_BW_Use"
+ },
+ {
+ "BriefDescription": "Average latency of data read request to external memory (in
nanoseconds). Accounts for demand loads and L1/L2 prefetches",
+ "MetricExpr": "1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD /
UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( cha_0@event\\=0x0@ /duration_time)",
+ "MetricGroup": "MemoryLat;SoC",
+ "MetricName": "MEM_Read_Latency"
+ },
+ {
+ "BriefDescription": "Average number of parallel data read requests to external memory.
Accounts for demand loads and L1/L2 prefetches",
+ "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD /
cha@event\\=0x36\\,umask\\=0xC817FE01\\,thresh\\=1@",
+ "MetricGroup": "MemoryBW;SoC",
+ "MetricName": "MEM_Parallel_Reads"
+ },
+ {
+ "BriefDescription": "Average latency of data read request to external 3D X-Point memory
[in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches",
+ "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM /
UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / cha_0@event\\=0x0@ )",
+ "MetricGroup": "MemoryLat;SoC;Server",
+ "MetricName": "MEM_PMM_Read_Latency"
+ },
+ {
+ "BriefDescription": "Average 3DXP Memory Bandwidth Use for reads [GB / sec]",
+ "MetricExpr": "( ( 64 * imc@event\\=0xe3@ / 1000000000 ) /duration_time)",
+ "MetricGroup": "MemoryBW;SoC;Server",
+ "MetricName": "PMM_Read_BW"
+ },
+ {
+ "BriefDescription": "Average 3DXP Memory Bandwidth Use for Writes [GB / sec]",
+ "MetricExpr": "( ( 64 * imc@event\\=0xe7@ / 1000000000 ) /duration_time)",
+ "MetricGroup": "MemoryBW;SoC;Server",
+ "MetricName": "PMM_Write_BW"
+ },
+ {
+ "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]",
+ "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000000 /duration_time",
+ "MetricGroup": "IoBW;SoC;Server",
+ "MetricName": "IO_Write_BW"
+ },
+ {
+ "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]",
+ "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM +
UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / 1000000000 /duration_time",
+ "MetricGroup": "IoBW;SoC;Server",
+ "MetricName": "IO_Read_BW"
+ },
+ {
+ "BriefDescription": "Socket actual clocks when any core is active on that socket",
+ "MetricExpr": "cha_0@event\\=0x0@",
+ "MetricGroup": "SoC",
+ "MetricName": "Socket_CLKS"
+ },
+ {
+ "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition
from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]",
+ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u",
+ "MetricGroup": "Branches;OS",
+ "MetricName": "IpFarBranch"
+ },
+ {
+ "BriefDescription": "C1 residency percent per core",
+ "MetricExpr": "(cstate_core@c1\\-residency@ / msr@tsc@) * 100",
+ "MetricGroup": "Power",
+ "MetricName": "C1_Core_Residency"
+ },
+ {
+ "BriefDescription": "C6 residency percent per core",
+ "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100",
+ "MetricGroup": "Power",
+ "MetricName": "C6_Core_Residency"
+ },
+ {
+ "BriefDescription": "C2 residency percent per package",
+ "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100",
+ "MetricGroup": "Power",
+ "MetricName": "C2_Pkg_Residency"
+ },
+ {
+ "BriefDescription": "C6 residency percent per package",
+ "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100",
+ "MetricGroup": "Power",
+ "MetricName": "C6_Pkg_Residency"
+ },
+]
Ian Rogers
reviewerQuoted 70 lines · Yao
Hi Andi,
I guess Arnaldo needs your ACK before consider picking up this patch.
Can you give an ACK for this patch?
Thanks
Jin YaoQuoted 59 lines · Yao
Hi Arnaldo,
Can you kindly pick up this patch?
Thanks
Jin YaoQuoted 51 lines · Jin Yao
Add JSON metrics for Icelake Server to perf.
Based on TMA metrics 4.21 at 01.org.
https://download.01.org/perfmon/Signed-off-by Jin Yao <[email protected]>---
v3:
- PMUcstate_coreandcstate_pkgare supported for ICX since 5.14-rc1, add cstate metrics for Core C1/C6 and Package C2/C6.v2:
- Fix perf test 10 error.# ./perf test 10 10: PMU events : 10.1: PMU event table sanity : Ok 10.2: PMU event map aliases : Ok 10.3: Parsing of PMU event table metrics : Ok 10.4: Parsing of PMU event table metrics with fake PMUs : Ok- Remove cstate metrics because the kernel has not supported
cstate_coreandcstate_corefor Icelake server.- Remove the topdown L1/L2 metrics.
.../arch/x86/icelakex/icx-metrics.json | 315 ++++++++++++++++++ 1 file changed, 315 insertions(+)create mode 100644
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.jsontools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +14
-0@@ -0,0 +1,315 @@ +[ + { + "BriefDescription": "Instructions Per Cycle (per Logical Processor)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Summary", + "MetricName": "IPC" + }, + { + "BriefDescription": "Uops Per Instruction", + "MetricExpr": "UOPS_RETIRED.SLOTS / INST_RETIRED.ANY", + "MetricGroup": "Pipeline;Retire", + "MetricName": "UPI" + }, + {
Quoted 8 lines
Quoted 8 lines
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json line 22
Quoted 8 lines
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +8
-0+ "BriefDescription": "Instruction per taken branch", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;FetchBW;PGO", + "MetricName": "IpTB" + }, + { + "BriefDescription": "Cycles Per Instruction (per Logical Processor)", + "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)",
Hi,
I'm not sure what's happening but when I generate this using:
https://github.com/intel/event-converter-for-linux-perf
I see here:
"MetricExpr": "1 / IPC",
There are various other differences and so I've included the generated file.
Thanks,
Ian
Quoted 335 lines
Quoted 335 lines
Quoted 335 lines
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "Pipeline", + "MetricName": "CPI" + }, + { + "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor isactive.",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +13
-0+ "MetricExpr": "CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "Pipeline", + "MetricName": "CLKS" + }, + { + "BriefDescription": "Instructions Per Cycle (per physical core)", + "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT;TmaL1", + "MetricName": "CoreIPC" + }, + { + "BriefDescription": "Floating Point Operations Per Cycle", + "MetricExpr": "( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / CPU_CLK_UNHALTED.DISTRIBUTED",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "Flops", + "MetricName": "FLOPc" + }, + { + "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed whenthere is at least 1 uop executed)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "UOPS_EXECUTED.THREAD / ( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 )", + "MetricGroup": "Pipeline;PortsUtil", + "MetricName": "ILP" + }, + { + "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction(JEClear)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", + "MetricGroup": "BrMispredicts", + "MetricName": "IpMispredict" + }, + { + "BriefDescription": "Core actual clocks when any Logical Processor is active on thePhysical Core",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +24
-0+ "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "CORE_CLKS" + }, + { + "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", + "MetricGroup": "InsType", + "MetricName": "IpLoad" + }, + { + "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", + "MetricGroup": "InsType", + "MetricName": "IpStore" + }, + { + "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", + "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", + "MetricGroup": "Branches;InsType", + "MetricName": "IpBranch" + }, + { + "BriefDescription": "Instructions per (near) call (lower number means higher occurrencerate)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +12
-0+ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", + "MetricGroup": "Branches", + "MetricName": "IpCall" + }, + { + "BriefDescription": "Branch instructions per taken branch. ", + "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", + "MetricGroup": "Branches;PGO", + "MetricName": "BpTkBranch" + }, + { + "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number meanshigher occurrence rate)",
+ "MetricExpr": "INST_RETIRED.ANY / ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +
FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "Flops;FpArith;InsType", + "MetricName": "IpFLOP" + }, + { + "BriefDescription": "Total number of retired Instructions, Sample with:INST_RETIRED.PREC_DIST",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "INST_RETIRED.ANY", + "MetricGroup": "Summary;TmaL1", + "MetricName": "Instructions" + }, + { + "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; akaLoop Cache)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "LSD", + "MetricName": "LSD_Coverage" + }, + { + "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or UopCache)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", + "MetricGroup": "DSB;FetchBW", + "MetricName": "DSB_Coverage" + }, + { + "BriefDescription": "Actual Average Latency for L1 data-cache miss demand loads (in corecycles)",
+ "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS +
MEM_LOAD_RETIRED.FB_HIT )",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "MemoryBound;MemoryLat", + "MetricName": "Load_Miss_Real_Latency" + }, + { + "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load whenthere is at least one such miss. Per-Logical Processor)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", + "MetricGroup": "MemoryBound;MemoryBW", + "MetricName": "MLP" + }, + { + "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB missestriggered by instruction/Load/Store accesses",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +2
-0+ "MetricConstraint": "NO_NMI_WATCHDOG", + "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING +DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +41
-0+ "MetricGroup": "MemoryTLB", + "MetricName": "Page_Walks_Utilization" + }, + { + "BriefDescription": "Average data fill bandwidth to the L1 data cache [GB / sec]", + "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L1D_Cache_Fill_BW" + }, + { + "BriefDescription": "Average data fill bandwidth to the L2 cache [GB / sec]", + "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L2_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW", + "MetricName": "L3_Cache_Fill_BW" + }, + { + "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", + "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time", + "MetricGroup": "MemoryBW;Offcore", + "MetricName": "L3_Cache_Access_BW" + }, + { + "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L1MPKI" + }, + { + "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L2MPKI" + }, + { + "BriefDescription": "L2 cache misses per kilo instruction for all request types(including speculative)",
+ "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD
) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / INST_RETIRED.ANY",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +11
-0+ "MetricGroup": "CacheMisses;Offcore", + "MetricName": "L2MPKI_All" + }, + { + "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", + "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", + "MetricGroup": "CacheMisses", + "MetricName": "L3MPKI" + }, + { + "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instructionwhere the evicted lines are dropped (no writeback to L3 or memory)",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +19
-0+ "MetricExpr": "1000 * L2_LINES_OUT.SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_Silent_PKI" + }, + { + "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", + "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / INST_RETIRED.ANY", + "MetricGroup": "L2Evicts;Server", + "MetricName": "L2_Evictions_NonSilent_PKI" + }, + { + "BriefDescription": "Average CPU Utilization", + "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@", + "MetricGroup": "HPC;Summary", + "MetricName": "CPU_Utilization" + }, + { + "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]", + "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ /1000000000 /
duration_time",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +6
-0+ "MetricGroup": "Summary;Power", + "MetricName": "Average_Frequency" + }, + { + "BriefDescription": "Giga Floating Point Operations Per Second", + "MetricExpr": "( ( 1 * ( FP_ARITH_INST_RETIRED.SCALAR_SINGLE +FP_ARITH_INST_RETIRED.SCALAR_DOUBLE ) + 2 * FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + 4 * ( FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE ) + 8 * ( FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE ) + 16 * FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE ) / 1000000000 ) /
duration_time",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +24
-0+ "MetricGroup": "Flops;HPC", + "MetricName": "GFLOPs" + }, + { + "BriefDescription": "Average Frequency Utilization relative nominal frequency", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC", + "MetricGroup": "Power", + "MetricName": "Turbo_Utilization" + }, + { + "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", + "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED", + "MetricGroup": "SMT", + "MetricName": "SMT_2T_Utilization" + }, + { + "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", + "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", + "MetricGroup": "OS", + "MetricName": "Kernel_Utilization" + }, + { + "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", + "MetricExpr": "( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) /1000000000 ) /
duration_time",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "HPC;MemoryBW;SoC", + "MetricName": "DRAM_BW_Use" + }, + { + "BriefDescription": "Average latency of data read request to external memory (innanoseconds). Accounts for demand loads and L1/L2 prefetches",
+ "MetricExpr": "1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD /
UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( cha_0@event\\=0x0@ /duration_time)",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "MemoryLat;SoC", + "MetricName": "MEM_Read_Latency" + }, + { + "BriefDescription": "Average number of parallel data read requests to external memory.Accounts for demand loads and L1/L2 prefetches",
+ "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD /
cha@event\\=0x36\\,umask\\=0xC817FE01\\,thresh\\=1@",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +5
-0+ "MetricGroup": "MemoryBW;SoC", + "MetricName": "MEM_Parallel_Reads" + }, + { + "BriefDescription": "Average latency of data read request to external 3D X-Point memory[in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches",
+ "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM /
UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / cha_0@event\\=0x0@ )",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +24
-0+ "MetricGroup": "MemoryLat;SoC;Server", + "MetricName": "MEM_PMM_Read_Latency" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for reads [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe3@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Read_BW" + }, + { + "BriefDescription": "Average 3DXP Memory Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "( ( 64 * imc@event\\=0xe7@ / 1000000000 ) / duration_time )", + "MetricGroup": "MemoryBW;SoC;Server", + "MetricName": "PMM_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", + "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000000 / duration_time", + "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Write_BW" + }, + { + "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", + "MetricExpr": "( UNC_CHA_TOR_INSERTS.IO_HIT_ITOM + UNC_CHA_TOR_INSERTS.IO_MISS_ITOM +UNC_CHA_TOR_INSERTS.IO_HIT_ITOMCACHENEAR + UNC_CHA_TOR_INSERTS.IO_MISS_ITOMCACHENEAR ) * 64 / 1000000000 /
duration_time",tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +11
-0+ "MetricGroup": "IoBW;SoC;Server", + "MetricName": "IO_Read_BW" + }, + { + "BriefDescription": "Socket actual clocks when any core is active on that socket", + "MetricExpr": "cha_0@event\\=0x0@", + "MetricGroup": "SoC", + "MetricName": "Socket_CLKS" + }, + { + "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transitionfrom application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]",
tools/perf/pmu-events/arch/x86/icelakex/icx-metrics.json new file +29
-0+ "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", + "MetricGroup": "Branches;OS", + "MetricName": "IpFarBranch" + }, + { + "BriefDescription": "C1 residency percent per core", + "MetricExpr": "(cstate_core@c1\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C1_Core_Residency" + }, + { + "BriefDescription": "C6 residency percent per core", + "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Core_Residency" + }, + { + "BriefDescription": "C2 residency percent per package", + "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C2_Pkg_Residency" + }, + { + "BriefDescription": "C6 residency percent per package", + "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100", + "MetricGroup": "Power", + "MetricName": "C6_Pkg_Residency" + }, +]
Hi Ian,
Quoted 5 lines · Ian Rogers
Hi,
I'm not sure what's happening but when I generate this using:
https://github.com/intel/event-converter-for-linux-perf
Quoted 7 lines
I see here:
"MetricExpr": "1 / IPC",There are various other differences and so I've included the generated file.
Thanks,
Ian
Please see following commit in event-converter-for-linux-perf repo.
commit b2265b953d57130a2ad8255d126bb69cf8552898
Author: Jin Yao <[email protected]> Date: Thu Jun 3 09:57:01 2021 +0800
Don't expand inner metric if possible
Before:
{
"MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", "BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricGroup": "Pipeline",
"MetricName": "CPI"
},
After:
{
"MetricExpr": "1 / IPC",
"BriefDescription": "Cycles Per Instruction (per Logical Processor)", "MetricGroup": "Pipeline",
"MetricName": "CPI"
},
We also provide a new option '--expr-events' for checking if it's a complex metric. For a simple metric, we probablly don't need to reuse the inner metric. For a complex metric, it's useful to reuse the inner metric.
I enable the option '--expr-events' when generating the metrics JSON. Since CPI is a simple metric, so the inner metric is still expanded.
For cstate metrics, only Core C1/C6 and Package C2/C6 are supported for ICX. The cstate metrics are appended manually right now.
Thanks
Jin Yao