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": "Uops Per Instruction", 22 "MetricExpr": "UOPS_RETIRED.SLOTS / INST_RETIRED.ANY", 23 "MetricGroup": "Pipeline;Ret;Retire", 24 "MetricName": "UPI" 25 }, 26 { 27 "BriefDescription": "Instruction per taken branch", 28 "MetricExpr": "UOPS_RETIRED.SLOTS / BR_INST_RETIRED.NEAR_TAKEN", 29 "MetricGroup": "Branches;Fed;FetchBW", 30 "MetricName": "UpTB" 31 }, 32 { 33 "BriefDescription": "Cycles Per Instruction (per Logical Processor)", 34 "MetricExpr": "1 / (INST_RETIRED.ANY / CPU_CLK_UNHALTED.THREAD)", 35 "MetricGroup": "Pipeline;Mem", 36 "MetricName": "CPI" 37 }, 38 { 39 "BriefDescription": "Per-Logical Processor actual clocks when the Logical Processor is active.", 40 "MetricExpr": "CPU_CLK_UNHALTED.THREAD", 41 "MetricGroup": "Pipeline", 42 "MetricName": "CLKS" 43 }, 44 { 45 "BriefDescription": "Total issue-pipeline slots (per-Physical Core till ICL; per-Logical Processor ICL onward)", 46 "MetricExpr": "TOPDOWN.SLOTS", 47 "MetricGroup": "TmaL1", 48 "MetricName": "SLOTS" 49 }, 50 { 51 "BriefDescription": "Fraction of Physical Core issue-slots utilized by this Logical Processor", 52 "MetricExpr": "TOPDOWN.SLOTS / ( TOPDOWN.SLOTS / 2 ) if #SMT_on else 1", 53 "MetricGroup": "SMT", 54 "MetricName": "Slots_Utilization" 55 }, 56 { 57 "BriefDescription": "The ratio of Executed- by Issued-Uops", 58 "MetricExpr": "UOPS_EXECUTED.THREAD / UOPS_ISSUED.ANY", 59 "MetricGroup": "Cor;Pipeline", 60 "MetricName": "Execute_per_Issue", 61 "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." 62 }, 63 { 64 "BriefDescription": "Instructions Per Cycle across hyper-threads (per physical core)", 65 "MetricExpr": "INST_RETIRED.ANY / CPU_CLK_UNHALTED.DISTRIBUTED", 66 "MetricGroup": "Ret;SMT;TmaL1", 67 "MetricName": "CoreIPC" 68 }, 69 { 70 "BriefDescription": "Floating Point Operations Per Cycle", 71 "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", 72 "MetricGroup": "Ret;Flops", 73 "MetricName": "FLOPc" 74 }, 75 { 76 "BriefDescription": "Actual per-core usage of the Floating Point execution units (regardless of the vector width)", 77 "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 )", 78 "MetricGroup": "Cor;Flops;HPC", 79 "MetricName": "FP_Arith_Utilization", 80 "PublicDescription": "Actual per-core usage of the Floating Point execution units (regardless of the vector width). Values > 1 are possible due to Fused-Multiply Add (FMA) counting." 81 }, 82 { 83 "BriefDescription": "Instruction-Level-Parallelism (average number of uops executed when there is at least 1 uop executed)", 84 "MetricExpr": "UOPS_EXECUTED.THREAD / (( UOPS_EXECUTED.CORE_CYCLES_GE_1 / 2 ) if #SMT_on else UOPS_EXECUTED.CORE_CYCLES_GE_1)", 85 "MetricGroup": "Backend;Cor;Pipeline;PortsUtil", 86 "MetricName": "ILP" 87 }, 88 { 89 "BriefDescription": "Number of Instructions per non-speculative Branch Misprediction (JEClear)", 90 "MetricExpr": "INST_RETIRED.ANY / BR_MISP_RETIRED.ALL_BRANCHES", 91 "MetricGroup": "Bad;BadSpec;BrMispredicts", 92 "MetricName": "IpMispredict" 93 }, 94 { 95 "BriefDescription": "Core actual clocks when any Logical Processor is active on the Physical Core", 96 "MetricExpr": "CPU_CLK_UNHALTED.DISTRIBUTED", 97 "MetricGroup": "SMT", 98 "MetricName": "CORE_CLKS" 99 }, 100 { 101 "BriefDescription": "Instructions per Load (lower number means higher occurrence rate)", 102 "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_LOADS", 103 "MetricGroup": "InsType", 104 "MetricName": "IpLoad" 105 }, 106 { 107 "BriefDescription": "Instructions per Store (lower number means higher occurrence rate)", 108 "MetricExpr": "INST_RETIRED.ANY / MEM_INST_RETIRED.ALL_STORES", 109 "MetricGroup": "InsType", 110 "MetricName": "IpStore" 111 }, 112 { 113 "BriefDescription": "Instructions per Branch (lower number means higher occurrence rate)", 114 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.ALL_BRANCHES", 115 "MetricGroup": "Branches;Fed;InsType", 116 "MetricName": "IpBranch" 117 }, 118 { 119 "BriefDescription": "Instructions per (near) call (lower number means higher occurrence rate)", 120 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_CALL", 121 "MetricGroup": "Branches;Fed;PGO", 122 "MetricName": "IpCall" 123 }, 124 { 125 "BriefDescription": "Instruction per taken branch", 126 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.NEAR_TAKEN", 127 "MetricGroup": "Branches;Fed;FetchBW;Frontend;PGO", 128 "MetricName": "IpTB" 129 }, 130 { 131 "BriefDescription": "Branch instructions per taken branch. ", 132 "MetricExpr": "BR_INST_RETIRED.ALL_BRANCHES / BR_INST_RETIRED.NEAR_TAKEN", 133 "MetricGroup": "Branches;Fed;PGO", 134 "MetricName": "BpTkBranch" 135 }, 136 { 137 "BriefDescription": "Instructions per Floating Point (FP) Operation (lower number means higher occurrence rate)", 138 "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 )", 139 "MetricGroup": "Flops;InsType", 140 "MetricName": "IpFLOP" 141 }, 142 { 143 "BriefDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate)", 144 "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) )", 145 "MetricGroup": "Flops;InsType", 146 "MetricName": "IpArith", 147 "PublicDescription": "Instructions per FP Arithmetic instruction (lower number means higher occurrence rate). May undercount due to FMA double counting. Approximated prior to BDW." 148 }, 149 { 150 "BriefDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate)", 151 "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_SINGLE", 152 "MetricGroup": "Flops;FpScalar;InsType", 153 "MetricName": "IpArith_Scalar_SP", 154 "PublicDescription": "Instructions per FP Arithmetic Scalar Single-Precision instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 155 }, 156 { 157 "BriefDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate)", 158 "MetricExpr": "INST_RETIRED.ANY / FP_ARITH_INST_RETIRED.SCALAR_DOUBLE", 159 "MetricGroup": "Flops;FpScalar;InsType", 160 "MetricName": "IpArith_Scalar_DP", 161 "PublicDescription": "Instructions per FP Arithmetic Scalar Double-Precision instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 162 }, 163 { 164 "BriefDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate)", 165 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.128B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.128B_PACKED_SINGLE )", 166 "MetricGroup": "Flops;FpVector;InsType", 167 "MetricName": "IpArith_AVX128", 168 "PublicDescription": "Instructions per FP Arithmetic AVX/SSE 128-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 169 }, 170 { 171 "BriefDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate)", 172 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.256B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.256B_PACKED_SINGLE )", 173 "MetricGroup": "Flops;FpVector;InsType", 174 "MetricName": "IpArith_AVX256", 175 "PublicDescription": "Instructions per FP Arithmetic AVX* 256-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 176 }, 177 { 178 "BriefDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate)", 179 "MetricExpr": "INST_RETIRED.ANY / ( FP_ARITH_INST_RETIRED.512B_PACKED_DOUBLE + FP_ARITH_INST_RETIRED.512B_PACKED_SINGLE )", 180 "MetricGroup": "Flops;FpVector;InsType", 181 "MetricName": "IpArith_AVX512", 182 "PublicDescription": "Instructions per FP Arithmetic AVX 512-bit instruction (lower number means higher occurrence rate). May undercount due to FMA double counting." 183 }, 184 { 185 "BriefDescription": "Total number of retired Instructions, Sample with: INST_RETIRED.PREC_DIST", 186 "MetricExpr": "INST_RETIRED.ANY", 187 "MetricGroup": "Summary;TmaL1", 188 "MetricName": "Instructions" 189 }, 190 { 191 "BriefDescription": "Average number of Uops issued by front-end when it issued something", 192 "MetricExpr": "UOPS_ISSUED.ANY / cpu@UOPS_ISSUED.ANY\\,cmask\\=1@", 193 "MetricGroup": "Fed;FetchBW", 194 "MetricName": "Fetch_UpC" 195 }, 196 { 197 "BriefDescription": "Fraction of Uops delivered by the LSD (Loop Stream Detector; aka Loop Cache)", 198 "MetricExpr": "LSD.UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", 199 "MetricGroup": "Fed;LSD", 200 "MetricName": "LSD_Coverage" 201 }, 202 { 203 "BriefDescription": "Fraction of Uops delivered by the DSB (aka Decoded ICache; or Uop Cache)", 204 "MetricExpr": "IDQ.DSB_UOPS / (IDQ.DSB_UOPS + LSD.UOPS + IDQ.MITE_UOPS + IDQ.MS_UOPS)", 205 "MetricGroup": "DSB;Fed;FetchBW", 206 "MetricName": "DSB_Coverage" 207 }, 208 { 209 "BriefDescription": "Number of Instructions per non-speculative DSB miss", 210 "MetricExpr": "INST_RETIRED.ANY / FRONTEND_RETIRED.ANY_DSB_MISS", 211 "MetricGroup": "DSBmiss;Fed", 212 "MetricName": "IpDSB_Miss_Ret" 213 }, 214 { 215 "BriefDescription": "Fraction of branches that are non-taken conditionals", 216 "MetricExpr": "BR_INST_RETIRED.COND_NTAKEN / BR_INST_RETIRED.ALL_BRANCHES", 217 "MetricGroup": "Bad;Branches;CodeGen;PGO", 218 "MetricName": "Cond_NT" 219 }, 220 { 221 "BriefDescription": "Fraction of branches that are taken conditionals", 222 "MetricExpr": "BR_INST_RETIRED.COND_TAKEN / BR_INST_RETIRED.ALL_BRANCHES", 223 "MetricGroup": "Bad;Branches;CodeGen;PGO", 224 "MetricName": "Cond_TK" 225 }, 226 { 227 "BriefDescription": "Fraction of branches that are CALL or RET", 228 "MetricExpr": "( BR_INST_RETIRED.NEAR_CALL + BR_INST_RETIRED.NEAR_RETURN ) / BR_INST_RETIRED.ALL_BRANCHES", 229 "MetricGroup": "Bad;Branches", 230 "MetricName": "CallRet" 231 }, 232 { 233 "BriefDescription": "Fraction of branches that are unconditional (direct or indirect) jumps", 234 "MetricExpr": "(BR_INST_RETIRED.NEAR_TAKEN - BR_INST_RETIRED.COND_TAKEN - 2 * BR_INST_RETIRED.NEAR_CALL) / BR_INST_RETIRED.ALL_BRANCHES", 235 "MetricGroup": "Bad;Branches", 236 "MetricName": "Jump" 237 }, 238 { 239 "BriefDescription": "Fraction of branches of other types (not individually covered by other metrics in Info.Branches group)", 240 "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) )", 241 "MetricGroup": "Bad;Branches", 242 "MetricName": "Other_Branches" 243 }, 244 { 245 "BriefDescription": "Actual Average Latency for L1 data-cache miss demand load instructions (in core cycles)", 246 "MetricExpr": "L1D_PEND_MISS.PENDING / ( MEM_LOAD_RETIRED.L1_MISS + MEM_LOAD_RETIRED.FB_HIT )", 247 "MetricGroup": "Mem;MemoryBound;MemoryLat", 248 "MetricName": "Load_Miss_Real_Latency", 249 "PublicDescription": "Actual Average Latency for L1 data-cache miss demand load instructions (in core cycles). Latency may be overestimated for multi-load instructions - e.g. repeat strings." 250 }, 251 { 252 "BriefDescription": "Memory-Level-Parallelism (average number of L1 miss demand load when there is at least one such miss. Per-Logical Processor)", 253 "MetricExpr": "L1D_PEND_MISS.PENDING / L1D_PEND_MISS.PENDING_CYCLES", 254 "MetricGroup": "Mem;MemoryBound;MemoryBW", 255 "MetricName": "MLP" 256 }, 257 { 258 "BriefDescription": "Average data fill bandwidth to the L1 data cache [GB / sec]", 259 "MetricExpr": "64 * L1D.REPLACEMENT / 1000000000 / duration_time", 260 "MetricGroup": "Mem;MemoryBW", 261 "MetricName": "L1D_Cache_Fill_BW" 262 }, 263 { 264 "BriefDescription": "Average data fill bandwidth to the L2 cache [GB / sec]", 265 "MetricExpr": "64 * L2_LINES_IN.ALL / 1000000000 / duration_time", 266 "MetricGroup": "Mem;MemoryBW", 267 "MetricName": "L2_Cache_Fill_BW" 268 }, 269 { 270 "BriefDescription": "Average per-core data fill bandwidth to the L3 cache [GB / sec]", 271 "MetricExpr": "64 * LONGEST_LAT_CACHE.MISS / 1000000000 / duration_time", 272 "MetricGroup": "Mem;MemoryBW", 273 "MetricName": "L3_Cache_Fill_BW" 274 }, 275 { 276 "BriefDescription": "Average per-core data access bandwidth to the L3 cache [GB / sec]", 277 "MetricExpr": "64 * OFFCORE_REQUESTS.ALL_REQUESTS / 1000000000 / duration_time", 278 "MetricGroup": "Mem;MemoryBW;Offcore", 279 "MetricName": "L3_Cache_Access_BW" 280 }, 281 { 282 "BriefDescription": "L1 cache true misses per kilo instruction for retired demand loads", 283 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L1_MISS / INST_RETIRED.ANY", 284 "MetricGroup": "Mem;CacheMisses", 285 "MetricName": "L1MPKI" 286 }, 287 { 288 "BriefDescription": "L1 cache true misses per kilo instruction for all demand loads (including speculative)", 289 "MetricExpr": "1000 * L2_RQSTS.ALL_DEMAND_DATA_RD / INST_RETIRED.ANY", 290 "MetricGroup": "Mem;CacheMisses", 291 "MetricName": "L1MPKI_Load" 292 }, 293 { 294 "BriefDescription": "L2 cache true misses per kilo instruction for retired demand loads", 295 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L2_MISS / INST_RETIRED.ANY", 296 "MetricGroup": "Mem;Backend;CacheMisses", 297 "MetricName": "L2MPKI" 298 }, 299 { 300 "BriefDescription": "L2 cache misses per kilo instruction for all request types (including speculative)", 301 "MetricExpr": "1000 * ( ( OFFCORE_REQUESTS.ALL_DATA_RD - OFFCORE_REQUESTS.DEMAND_DATA_RD ) + L2_RQSTS.ALL_DEMAND_MISS + L2_RQSTS.SWPF_MISS ) / INST_RETIRED.ANY", 302 "MetricGroup": "Mem;CacheMisses;Offcore", 303 "MetricName": "L2MPKI_All" 304 }, 305 { 306 "BriefDescription": "L2 cache misses per kilo instruction for all demand loads (including speculative)", 307 "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_MISS / INST_RETIRED.ANY", 308 "MetricGroup": "Mem;CacheMisses", 309 "MetricName": "L2MPKI_Load" 310 }, 311 { 312 "BriefDescription": "L2 cache hits per kilo instruction for all demand loads (including speculative)", 313 "MetricExpr": "1000 * L2_RQSTS.DEMAND_DATA_RD_HIT / INST_RETIRED.ANY", 314 "MetricGroup": "Mem;CacheMisses", 315 "MetricName": "L2HPKI_Load" 316 }, 317 { 318 "BriefDescription": "L3 cache true misses per kilo instruction for retired demand loads", 319 "MetricExpr": "1000 * MEM_LOAD_RETIRED.L3_MISS / INST_RETIRED.ANY", 320 "MetricGroup": "Mem;CacheMisses", 321 "MetricName": "L3MPKI" 322 }, 323 { 324 "BriefDescription": "Fill Buffer (FB) true hits per kilo instructions for retired demand loads", 325 "MetricExpr": "1000 * MEM_LOAD_RETIRED.FB_HIT / INST_RETIRED.ANY", 326 "MetricGroup": "Mem;CacheMisses", 327 "MetricName": "FB_HPKI" 328 }, 329 { 330 "BriefDescription": "Utilization of the core's Page Walker(s) serving STLB misses triggered by instruction/Load/Store accesses", 331 "MetricConstraint": "NO_NMI_WATCHDOG", 332 "MetricExpr": "( ITLB_MISSES.WALK_PENDING + DTLB_LOAD_MISSES.WALK_PENDING + DTLB_STORE_MISSES.WALK_PENDING ) / ( 2 * CPU_CLK_UNHALTED.DISTRIBUTED )", 333 "MetricGroup": "Mem;MemoryTLB", 334 "MetricName": "Page_Walks_Utilization" 335 }, 336 { 337 "BriefDescription": "Rate of silent evictions from the L2 cache per Kilo instruction where the evicted lines are dropped (no writeback to L3 or memory)", 338 "MetricExpr": "1000 * L2_LINES_OUT.SILENT / INST_RETIRED.ANY", 339 "MetricGroup": "L2Evicts;Mem;Server", 340 "MetricName": "L2_Evictions_Silent_PKI" 341 }, 342 { 343 "BriefDescription": "Rate of non silent evictions from the L2 cache per Kilo instruction", 344 "MetricExpr": "1000 * L2_LINES_OUT.NON_SILENT / INST_RETIRED.ANY", 345 "MetricGroup": "L2Evicts;Mem;Server", 346 "MetricName": "L2_Evictions_NonSilent_PKI" 347 }, 348 { 349 "BriefDescription": "Average CPU Utilization", 350 "MetricExpr": "CPU_CLK_UNHALTED.REF_TSC / msr@tsc@", 351 "MetricGroup": "HPC;Summary", 352 "MetricName": "CPU_Utilization" 353 }, 354 { 355 "BriefDescription": "Measured Average Frequency for unhalted processors [GHz]", 356 "MetricExpr": "(CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC) * msr@tsc@ / 1000000000 / duration_time", 357 "MetricGroup": "Summary;Power", 358 "MetricName": "Average_Frequency" 359 }, 360 { 361 "BriefDescription": "Giga Floating Point Operations Per Second", 362 "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", 363 "MetricGroup": "Cor;Flops;HPC", 364 "MetricName": "GFLOPs" 365 }, 366 { 367 "BriefDescription": "Average Frequency Utilization relative nominal frequency", 368 "MetricExpr": "CPU_CLK_UNHALTED.THREAD / CPU_CLK_UNHALTED.REF_TSC", 369 "MetricGroup": "Power", 370 "MetricName": "Turbo_Utilization" 371 }, 372 { 373 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for baseline license level 0", 374 "MetricExpr": "CORE_POWER.LVL0_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 375 "MetricGroup": "Power", 376 "MetricName": "Power_License0_Utilization", 377 "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." 378 }, 379 { 380 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 1", 381 "MetricExpr": "CORE_POWER.LVL1_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 382 "MetricGroup": "Power", 383 "MetricName": "Power_License1_Utilization", 384 "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." 385 }, 386 { 387 "BriefDescription": "Fraction of Core cycles where the core was running with power-delivery for license level 2 (introduced in SKX)", 388 "MetricExpr": "CORE_POWER.LVL2_TURBO_LICENSE / CPU_CLK_UNHALTED.DISTRIBUTED", 389 "MetricGroup": "Power", 390 "MetricName": "Power_License2_Utilization", 391 "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." 392 }, 393 { 394 "BriefDescription": "Fraction of cycles where both hardware Logical Processors were active", 395 "MetricExpr": "1 - CPU_CLK_UNHALTED.ONE_THREAD_ACTIVE / CPU_CLK_UNHALTED.REF_DISTRIBUTED if #SMT_on else 0", 396 "MetricGroup": "SMT", 397 "MetricName": "SMT_2T_Utilization" 398 }, 399 { 400 "BriefDescription": "Fraction of cycles spent in the Operating System (OS) Kernel mode", 401 "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / CPU_CLK_UNHALTED.THREAD", 402 "MetricGroup": "OS", 403 "MetricName": "Kernel_Utilization" 404 }, 405 { 406 "BriefDescription": "Cycles Per Instruction for the Operating System (OS) Kernel mode", 407 "MetricExpr": "CPU_CLK_UNHALTED.THREAD_P:k / INST_RETIRED.ANY_P:k", 408 "MetricGroup": "OS", 409 "MetricName": "Kernel_CPI" 410 }, 411 { 412 "BriefDescription": "Average external Memory Bandwidth Use for reads and writes [GB / sec]", 413 "MetricExpr": "( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) / 1000000000 ) / duration_time", 414 "MetricGroup": "HPC;Mem;MemoryBW;SoC", 415 "MetricName": "DRAM_BW_Use" 416 }, 417 { 418 "BriefDescription": "Average latency of data read request to external memory (in nanoseconds). Accounts for demand loads and L1/L2 prefetches", 419 "MetricExpr": "1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / UNC_CHA_TOR_INSERTS.IA_MISS_DRD ) / ( cha_0@event\\=0x0@ / duration_time )", 420 "MetricGroup": "Mem;MemoryLat;SoC", 421 "MetricName": "MEM_Read_Latency" 422 }, 423 { 424 "BriefDescription": "Average number of parallel data read requests to external memory. Accounts for demand loads and L1/L2 prefetches", 425 "MetricExpr": "UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD / cha@event\\=0x36\\,umask\\=0xC817FE01\\,thresh\\=1@", 426 "MetricGroup": "Mem;MemoryBW;SoC", 427 "MetricName": "MEM_Parallel_Reads" 428 }, 429 { 430 "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", 431 "MetricExpr": "( 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_PMM / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_PMM ) / cha_0@event\\=0x0@ )", 432 "MetricGroup": "Mem;MemoryLat;SoC;Server", 433 "MetricName": "MEM_PMM_Read_Latency" 434 }, 435 { 436 "BriefDescription": "Average latency of data read request to external DRAM memory [in nanoseconds]. Accounts for demand loads and L1/L2 data-read prefetches", 437 "MetricExpr": " 1000000000 * ( UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_DDR / UNC_CHA_TOR_INSERTS.IA_MISS_DRD_DDR ) / cha_0@event\\=0x0@", 438 "MetricGroup": "Mem;MemoryLat;SoC;Server", 439 "MetricName": "MEM_DRAM_Read_Latency" 440 }, 441 { 442 "BriefDescription": "Average 3DXP Memory Bandwidth Use for reads [GB / sec]", 443 "MetricExpr": "( ( 64 * imc@event\\=0xe3@ / 1000000000 ) / duration_time )", 444 "MetricGroup": "Mem;MemoryBW;SoC;Server", 445 "MetricName": "PMM_Read_BW" 446 }, 447 { 448 "BriefDescription": "Average 3DXP Memory Bandwidth Use for Writes [GB / sec]", 449 "MetricExpr": "( ( 64 * imc@event\\=0xe7@ / 1000000000 ) / duration_time )", 450 "MetricGroup": "Mem;MemoryBW;SoC;Server", 451 "MetricName": "PMM_Write_BW" 452 }, 453 { 454 "BriefDescription": "Average IO (network or disk) Bandwidth Use for Writes [GB / sec]", 455 "MetricExpr": "UNC_CHA_TOR_INSERTS.IO_PCIRDCUR * 64 / 1000000000 / duration_time", 456 "MetricGroup": "IoBW;Mem;SoC;Server", 457 "MetricName": "IO_Write_BW" 458 }, 459 { 460 "BriefDescription": "Average IO (network or disk) Bandwidth Use for Reads [GB / sec]", 461 "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", 462 "MetricGroup": "IoBW;Mem;SoC;Server", 463 "MetricName": "IO_Read_BW" 464 }, 465 { 466 "BriefDescription": "Socket actual clocks when any core is active on that socket", 467 "MetricExpr": "cha_0@event\\=0x0@", 468 "MetricGroup": "SoC", 469 "MetricName": "Socket_CLKS" 470 }, 471 { 472 "BriefDescription": "Instructions per Far Branch ( Far Branches apply upon transition from application to operating system, handling interrupts, exceptions) [lower number means higher occurrence rate]", 473 "MetricExpr": "INST_RETIRED.ANY / BR_INST_RETIRED.FAR_BRANCH:u", 474 "MetricGroup": "Branches;OS", 475 "MetricName": "IpFarBranch" 476 }, 477 { 478 "BriefDescription": "C1 residency percent per core", 479 "MetricExpr": "(cstate_core@c1\\-residency@ / msr@tsc@) * 100", 480 "MetricGroup": "Power", 481 "MetricName": "C1_Core_Residency" 482 }, 483 { 484 "BriefDescription": "C6 residency percent per core", 485 "MetricExpr": "(cstate_core@c6\\-residency@ / msr@tsc@) * 100", 486 "MetricGroup": "Power", 487 "MetricName": "C6_Core_Residency" 488 }, 489 { 490 "BriefDescription": "C2 residency percent per package", 491 "MetricExpr": "(cstate_pkg@c2\\-residency@ / msr@tsc@) * 100", 492 "MetricGroup": "Power", 493 "MetricName": "C2_Pkg_Residency" 494 }, 495 { 496 "BriefDescription": "C6 residency percent per package", 497 "MetricExpr": "(cstate_pkg@c6\\-residency@ / msr@tsc@) * 100", 498 "MetricGroup": "Power", 499 "MetricName": "C6_Pkg_Residency" 500 } 501] 502