1[ 2 { 3 "BriefDescription": "Total pipeline cost of branch related instructions (used for program control-flow including function calls)", 4 "MetricExpr": "100 * (( BR_INST_RETIRED.COND + 3 * BR_INST_RETIRED.NEAR_CALL + (BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL) ) / TOPDOWN.SLOTS)", 5 "MetricGroup": "Ret", 6 "MetricName": "Branching_Overhead" 7 }, 8 { 9 "BriefDescription": "Total pipeline cost of instruction fetch related bottlenecks by large code footprint programs (i-side cache; TLB and BTB misses)", 10 "MetricExpr": "100 * (( 5 * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / TOPDOWN.SLOTS) * ( (ICACHE_64B.IFTAG_STALL / CPU_CLK_UNHALTED.THREAD) + (ICACHE_16B.IFDATA_STALL / CPU_CLK_UNHALTED.THREAD) + (10 * BACLEARS.ANY / CPU_CLK_UNHALTED.THREAD) ) / #(( 5 * IDQ_UOPS_NOT_DELIVERED.CYCLES_0_UOPS_DELIV.CORE - INT_MISC.UOP_DROPPING ) / TOPDOWN.SLOTS)", 11 "MetricGroup": "BigFoot;Fed;Frontend;IcMiss;MemoryTLB", 12 "MetricName": "Big_Code" 13 }, 14 { 15 "BriefDescription": "Instructions Per Cycle (per Logical Processor)", 16 "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD", 17 "MetricGroup": "Ret;Summary", 18 "MetricName": "IPC" 19 }, 20 { 21 "BriefDescription": "Cycles Per Instruction (per Logical Processor)", 22 "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", 23 "MetricGroup": "Pipeline;Mem", 24 "MetricName": "CPI" 25 }, 26 { 27 "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", 28 "MetricExpr": "CPU_CLK_UNHALTED.THREAD", 29 "MetricGroup": "Pipeline", 30 "MetricName": "CLKS" 31 }, 32 { 33 "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", 34 "MetricExpr": "TOPDOWN.SLOTS", 35 "MetricGroup": "TmaL1", 36 "MetricName": "SLOTS" 37 }, 38 { 39 "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", 40 "MetricExpr": "TOPDOWN.SLOTS / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", 41 "MetricGroup": "SMT;TmaL1", 42 "MetricName": "Slots_Utilization" 43 }, 44 { 45 "BriefDescription": "The ratio of Executed- by Issued-Uops", 46 "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", 47 "MetricGroup": "Cor;Pipeline", 48 "MetricName": "Execute_per_Issue", 49 "PublicDescription": "The ratio of Executed- by Issued-Uops. Ratio > 1 suggests high rate of uop micro-fusions. Ratio < 1 suggest high rate of \"execute\" at rename stage." 50 }, 51 { 52 "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", 53 "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED", 54 "MetricGroup": "Ret;SMT;TmaL1", 55 "MetricName": "CoreIPC" 56 }, 57 { 58 "BriefDescription": "Floating Point Operations Per Cycle", 59 "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", 60 "MetricGroup": "Ret;Flops", 61 "MetricName": "FLOPc" 62 }, 63 { 64 "BriefDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width)", 65 "MetricExpr": "( (FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE) + (FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE) ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )", 66 "MetricGroup": "Cor;Flops;HPC", 67 "MetricName": "FP_Arith_Utilization", 68 "PublicDescription": "Actual per-core usage of the Floating Point non-X87 execution units (regardless of precision or vector-width). Values > 1 are possible due to ([BDW+] Fused-Multiply Add (FMA) counting - common; [ADL+] use all of ADD/MUL/FMA in Scalar or 128/256-bit vectors - less common)." 69 }, 70 { 71 "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is execution) per-core", 72 "MetricExpr": "UOPS_EXECUTED.THREAD / (( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else UOPS_EXECUTED.CORE_CYCLES_GE_1)", 73 "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", 74 "MetricName": "ILP" 75 }, 76 { 77 "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", 78 "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED", 79 "MetricGroup": "SMT", 80 "MetricName": "CORE_CLKS" 81 }, 82 { 83 "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", 84 "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", 85 "MetricGroup": "InsType", 86 "MetricName": "IpLoad" 87 }, 88 { 89 "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", 90 "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", 91 "MetricGroup": "InsType", 92 "MetricName": "IpStore" 93 }, 94 { 95 "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", 96 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", 97 "MetricGroup": "Branches;Fed;InsType", 98 "MetricName": "IpBranch" 99 }, 100 { 101 "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", 102 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", 103 "MetricGroup": "Branches;Fed;PGO", 104 "MetricName": "IpCall" 105 }, 106 { 107 "BriefDescription": "Instruction per taken branch", 108 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", 109 "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", 110 "MetricName": "IpTB" 111 }, 112 { 113 "BriefDescription": "Branch instructions per taken branch. ", 114 "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", 115 "MetricGroup": "Branches;Fed;PGO", 116 "MetricName": "BpTkBranch" 117 }, 118 { 119 "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", 120 "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 )", 121 "MetricGroup": "Flops;InsType", 122 "MetricName": "IpFLOP" 123 }, 124 { 125 "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", 126 "MetricExpr": "INST_RETIRED.ANY / ( (FP_ARITH_INST_RETIRED.SCALAR_SINGLE + FP_ARITH_INST_RETIRED.SCALAR_DOUBLE) + (FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE + FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE) )", 127 "MetricGroup": "Flops;InsType", 128 "MetricName": "IpArith", 129 "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). May undercount due to FMA double counting. Approximated prior to BDW." 130 }, 131 { 132 "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", 133 "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", 134 "MetricGroup": "Flops;FpScalar;InsType", 135 "MetricName": "IpArith_Scalar_SP", 136 "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 137 }, 138 { 139 "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", 140 "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", 141 "MetricGroup": "Flops;FpScalar;InsType", 142 "MetricName": "IpArith_Scalar_DP", 143 "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 144 }, 145 { 146 "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", 147 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", 148 "MetricGroup": "Flops;FpVector;InsType", 149 "MetricName": "IpArith_AVX128", 150 "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 151 }, 152 { 153 "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", 154 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", 155 "MetricGroup": "Flops;FpVector;InsType", 156 "MetricName": "IpArith_AVX256", 157 "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 158 }, 159 { 160 "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", 161 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", 162 "MetricGroup": "Flops;FpVector;InsType", 163 "MetricName": "IpArith_AVX512", 164 "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 165 }, 166 { 167 "BriefDescription": "Instructions per Software prefetch instruction (of any type: NTA/T0/T1/T2/Prefetch) (lower number means higher occurrence rate)", 168 "MetricExpr": "INST_RETIRED.ANY / cpu@SW_PREFETCH_ACCESS.T0\\,umask\\=0xF@", 169 "MetricGroup": "Prefetches", 170 "MetricName": "IpSWPF" 171 }, 172 { 173 "BriefDescription": "Total number of retired Instructions, Sample with: INST_RETIRED.PREC_DIST", 174 "MetricExpr": "INST_RETIRED.ANY", 175 "MetricGroup": "Summary;TmaL1", 176 "MetricName": "Instructions" 177 }, 178 { 179 "BriefDescription": "", 180 "MetricExpr": "UOPS_EXECUTED.THREAD / cpu@UOPS_EXECUTED.THREAD\\,cmask\\=1@", 181 "MetricGroup": "Cor;Pipeline;PortsUtil;SMT", 182 "MetricName": "Execute" 183 }, 184 { 185 "BriefDescription": "Average number of Uops issued by front-end when it issued something", 186 "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=1@", 187 "MetricGroup": "Fed;FetchBW", 188 "MetricName": "Fetch_UpC" 189 }, 190 { 191 "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", 192 "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", 193 "MetricGroup": "Fed;LSD", 194 "MetricName": "LSD_Coverage" 195 }, 196 { 197 "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", 198 "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", 199 "MetricGroup": "DSB;Fed;FetchBW", 200 "MetricName": "DSB_Coverage" 201 }, 202 { 203 "BriefDescription": "Average number of cycles of a switch from the DSB fetch-unit to MITE fetch unit - see DSB_Switches tree node for details.", 204 "MetricExpr": "DSB2MITE_SWITCHES.PENALTY_CYCLES / cpu@DSB2MITE_SWITCHES.PENALTY_CYCLES\\,cmask\\=1\\,edge@", 205 "MetricGroup": "DSBmiss", 206 "MetricName": "DSB_Switch_Cost" 207 }, 208 { 209 "BriefDescription": "Number of Instructions per non-speculative DSB miss (lower number means higher occurrence rate)", 210 "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", 211 "MetricGroup": "DSBmiss;Fed", 212 "MetricName": "IpDSB_Miss_Ret" 213 }, 214 { 215 "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear) (lower number means higher occurrence rate)", 216 "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", 217 "MetricGroup": "Bad;BadSpec;BrMispredicts", 218 "MetricName": "IpMispredict" 219 }, 220 { 221 "BriefDescription": "Fraction of branches that are non-taken conditionals", 222 "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", 223 "MetricGroup": "Bad;Branches;CodeGen;PGO", 224 "MetricName": "Cond_NT" 225 }, 226 { 227 "BriefDescription": "Fraction of branches that are taken conditionals", 228 "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", 229 "MetricGroup": "Bad;Branches;CodeGen;PGO", 230 "MetricName": "Cond_TK" 231 }, 232 { 233 "BriefDescription": "Fraction of branches that are CALL or RET", 234 "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", 235 "MetricGroup": "Bad;Branches", 236 "MetricName": "CallRet" 237 }, 238 { 239 "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", 240 "MetricExpr": "(BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL) / BR_INST_RETIRED.ALL_BRANCHES", 241 "MetricGroup": "Bad;Branches", 242 "MetricName": "Jump" 243 }, 244 { 245 "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", 246 "MetricExpr": "1 - ( (BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES) + (BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES) + (( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES) + ((BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL) / BR_INST_RETIRED.ALL_BRANCHES) )", 247 "MetricGroup": "Bad;Branches", 248 "MetricName": "Other_Branches" 249 }, 250 { 251 "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load operations (in core cycles)", 252 "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", 253 "MetricGroup": "Mem;MemoryBound;MemoryLat", 254 "MetricName": "Load_Miss_Real_Latency" 255 }, 256 { 257 "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)", 258 "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", 259 "MetricGroup": "Mem;MemoryBound;MemoryBW", 260 "MetricName": "MLP" 261 }, 262 { 263 "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", 264 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", 265 "MetricGroup": "Mem;CacheMisses", 266 "MetricName": "L1MPKI" 267 }, 268 { 269 "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", 270 "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", 271 "MetricGroup": "Mem;CacheMisses", 272 "MetricName": "L1MPKI_Load" 273 }, 274 { 275 "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", 276 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", 277 "MetricGroup": "Mem;Backend;CacheMisses", 278 "MetricName": "L2MPKI" 279 }, 280 { 281 "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all request types (including speculative)", 282 "MetricExpr": "1000 * L2_RQSTS.MISS / INST_RETIRED.ANY", 283 "MetricGroup": "Mem;CacheMisses;Offcore", 284 "MetricName": "L2MPKI_All" 285 }, 286 { 287 "BriefDescription": "L2 cache ([RKL+] true) misses per kilo instruction for all demand loads (including speculative)", 288 "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", 289 "MetricGroup": "Mem;CacheMisses", 290 "MetricName": "L2MPKI_Load" 291 }, 292 { 293 "BriefDescription": "L2 cache hits per kilo instruction for all request types (including speculative)", 294 "MetricExpr": "1000 * ( L2_RQSTS.REFERENCES - L2_RQSTS.MISS ) / INST_RETIRED.ANY", 295 "MetricGroup": "Mem;CacheMisses", 296 "MetricName": "L2HPKI_All" 297 }, 298 { 299 "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", 300 "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", 301 "MetricGroup": "Mem;CacheMisses", 302 "MetricName": "L2HPKI_Load" 303 }, 304 { 305 "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", 306 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", 307 "MetricGroup": "Mem;CacheMisses", 308 "MetricName": "L3MPKI" 309 }, 310 { 311 "BriefDescription": "Fill Buffer (FB) hits per kilo instructions for retired demand loads (L1D misses that merge into ongoing miss-handling entries)", 312 "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", 313 "MetricGroup": "Mem;CacheMisses", 314 "MetricName": "FB_HPKI" 315 }, 316 { 317 "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", 318 "MetricConstraint": "NO_NMI_WATCHDOG", 319 "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )", 320 "MetricGroup": "Mem;MemoryTLB", 321 "MetricName": "Page_Walks_Utilization" 322 }, 323 { 324 "BriefDescription": "Average per-core data fill bandwidth to the L1 data cache [GB / sec]", 325 "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 / duration_time", 326 "MetricGroup": "Mem;MemoryBW", 327 "MetricName": "L1D_Cache_Fill_BW" 328 }, 329 { 330 "BriefDescription": "Average per-core data fill bandwidth to the L2 cache [GB / sec]", 331 "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 / duration_time", 332 "MetricGroup": "Mem;MemoryBW", 333 "MetricName": "L2_Cache_Fill_BW" 334 }, 335 { 336 "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", 337 "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time", 338 "MetricGroup": "Mem;MemoryBW", 339 "MetricName": "L3_Cache_Fill_BW" 340 }, 341 { 342 "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", 343 "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time", 344 "MetricGroup": "Mem;MemoryBW;Offcore", 345 "MetricName": "L3_Cache_Access_BW" 346 }, 347 { 348 "BriefDescription": "Average per-thread data fill bandwidth to the L1 data cache [GB / sec]", 349 "MetricExpr": "(64 * L1D.REPLACEMENT / 1000000000 / duration_time)", 350 "MetricGroup": "Mem;MemoryBW", 351 "MetricName": "L1D_Cache_Fill_BW_1T" 352 }, 353 { 354 "BriefDescription": "Average per-thread data fill bandwidth to the L2 cache [GB / sec]", 355 "MetricExpr": "(64 * L2_LINES_IN.ALL / 1000000000 / duration_time)", 356 "MetricGroup": "Mem;MemoryBW", 357 "MetricName": "L2_Cache_Fill_BW_1T" 358 }, 359 { 360 "BriefDescription": "Average per-thread data fill bandwidth to the L3 cache [GB / sec]", 361 "MetricExpr": "(64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time)", 362 "MetricGroup": "Mem;MemoryBW", 363 "MetricName": "L3_Cache_Fill_BW_1T" 364 }, 365 { 366 "BriefDescription": "Average per-thread data access bandwidth to the L3 cache [GB / sec]", 367 "MetricExpr": "(64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time)", 368 "MetricGroup": "Mem;MemoryBW;Offcore", 369 "MetricName": "L3_Cache_Access_BW_1T" 370 }, 371 { 372 "BriefDescription": "Average CPU Utilization", 373 "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@", 374 "MetricGroup": "HPC;Summary", 375 "MetricName": "CPU_Utilization" 376 }, 377 { 378 "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]", 379 "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ / 1000000000 / duration_time", 380 "MetricGroup": "Summary;Power", 381 "MetricName": "Average_Frequency" 382 }, 383 { 384 "BriefDescription": "Giga Floating Point Operations Per Second", 385 "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", 386 "MetricGroup": "Cor;Flops;HPC", 387 "MetricName": "GFLOPs", 388 "PublicDescription": "Giga Floating Point Operations Per Second. Aggregate across all supported options of: FP precisions, scalar and vector instructions, vector-width and AMX engine." 389 }, 390 { 391 "BriefDescription": "Average Frequency Utilization relative nominal frequency", 392 "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC", 393 "MetricGroup": "Power", 394 "MetricName": "Turbo_Utilization" 395 }, 396 { 397 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", 398 "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 399 "MetricGroup": "Power", 400 "MetricName": "Power_License0_Utilization", 401 "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0. This includes non-AVX codes, SSE, AVX 128-bit, and low-current AVX 256-bit codes." 402 }, 403 { 404 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", 405 "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 406 "MetricGroup": "Power", 407 "MetricName": "Power_License1_Utilization", 408 "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1. This includes high current AVX 256-bit instructions as well as low current AVX 512-bit instructions." 409 }, 410 { 411 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", 412 "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 413 "MetricGroup": "Power", 414 "MetricName": "Power_License2_Utilization", 415 "PublicDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX). This includes high current AVX 512-bit instructions." 416 }, 417 { 418 "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", 419 "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", 420 "MetricGroup": "SMT", 421 "MetricName": "SMT_2T_Utilization" 422 }, 423 { 424 "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", 425 "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", 426 "MetricGroup": "OS", 427 "MetricName": "Kernel_Utilization" 428 }, 429 { 430 "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", 431 "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", 432 "MetricGroup": "OS", 433 "MetricName": "Kernel_CPI" 434 }, 435 { 436 "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", 437 "MetricExpr": "64 * ( arb@event\\=0x81\\,umask\\=0x1@ + arb@event\\=0x84\\,umask\\=0x1@ ) / 1000000 / duration_time / 1000", 438 "MetricGroup": "HPC;Mem;MemoryBW;SoC", 439 "MetricName": "DRAM_BW_Use" 440 }, 441 { 442 "BriefDescription": "Average number of parallel requests to external memory. Accounts for all requests", 443 "MetricExpr": "UNC_ARB_TRK_OCCUPANCY.ALL / arb@event\\=0x81\\,umask\\=0x1@", 444 "MetricGroup": "Mem;SoC", 445 "MetricName": "MEM_Parallel_Requests" 446 }, 447 { 448 "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", 449 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", 450 "MetricGroup": "Branches;OS", 451 "MetricName": "IpFarBranch" 452 }, 453 { 454 "BriefDescription": "C6 residency percent per core", 455 "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100", 456 "MetricGroup": "Power", 457 "MetricName": "C6_Core_Residency" 458 }, 459 { 460 "BriefDescription": "C7 residency percent per core", 461 "MetricExpr": "(cstate_core@c7\\-residency@ / msr@tsc@) * 100", 462 "MetricGroup": "Power", 463 "MetricName": "C7_Core_Residency" 464 }, 465 { 466 "BriefDescription": "C2 residency percent per package", 467 "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100", 468 "MetricGroup": "Power", 469 "MetricName": "C2_Pkg_Residency" 470 }, 471 { 472 "BriefDescription": "C3 residency percent per package", 473 "MetricExpr": "(cstate_pkg@c3\\-residency@ / msr@tsc@) * 100", 474 "MetricGroup": "Power", 475 "MetricName": "C3_Pkg_Residency" 476 }, 477 { 478 "BriefDescription": "C6 residency percent per package", 479 "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100", 480 "MetricGroup": "Power", 481 "MetricName": "C6_Pkg_Residency" 482 }, 483 { 484 "BriefDescription": "C7 residency percent per package", 485 "MetricExpr": "(cstate_pkg@c7\\-residency@ / msr@tsc@) * 100", 486 "MetricGroup": "Power", 487 "MetricName": "C7_Pkg_Residency" 488 }, 489 { 490 "BriefDescription": "C8 residency percent per package", 491 "MetricExpr": "(cstate_pkg@c8\\-residency@ / msr@tsc@) * 100", 492 "MetricGroup": "Power", 493 "MetricName": "C8_Pkg_Residency" 494 }, 495 { 496 "BriefDescription": "C9 residency percent per package", 497 "MetricExpr": "(cstate_pkg@c9\\-residency@ / msr@tsc@) * 100", 498 "MetricGroup": "Power", 499 "MetricName": "C9_Pkg_Residency" 500 }, 501 { 502 "BriefDescription": "C10 residency percent per package", 503 "MetricExpr": "(cstate_pkg@c10\\-residency@ / msr@tsc@) * 100", 504 "MetricGroup": "Power", 505 "MetricName": "C10_Pkg_Residency" 506 } 507] 508