파일
rocm-systems/tests/workloads/dev01p3/mi100/prev_analysis/201.csv
T
colramos-amd 62d130b458 Initial commit
2022-11-04 14:49:36 -05:00

1.5 KiB

1MetricValueUnitPeakPoP
2VALU FLOPsGflops23070.72
3VALU IOPsGops23070.72
4MFMA FLOPs (BF16)Gflops92282.88
5MFMA FLOPs (F16)Gflops184565.76
6MFMA FLOPs (F32)Gflops46141.44
7MFMA FLOPs (F64)Gflops46141.44
8MFMA IOPs (Int8)Gops184565.76
9Active CUs89Cus12074.16666666666667
10SALU Util2.1142315362209505Pct1002.1142315362209505
11VALU Util59.27505815065561Pct10059.27505815065561
12MFMA UtilPct100
13VALU Active Threads/Wave63.96805399754944Threads6499.950084371171
14IPC - Issue0.8437287060051507Instr/cycle516.874574120103013
15LDS BW0.0Gb/sec23070.720.0
16LDS Bank ConflictConflicts/access32
17Instr Cache Hit Rate99.99351101937752Pct10099.99351101937752
18Instr Cache BW1410.0001369988809Gb/s4614.14430.558217017043262
19Scalar L1D Cache Hit Rate99.3562044853034Pct10099.3562044853034
20Scalar L1D Cache BW43.047065285053684Gb/s4614.1440.9329371880256377
21Vector L1D Cache Hit Rate20.508982035928145Pct10020.508982035928145
22Vector L1D Cache BW1219.6973641978805Gb/s11535.3610.573552660670153
23L2 Cache Hit Rate0.3929601391220619Pct1000.3929601391220619
24L2-Fabric Read BW837.9967513054252Gb/s1228.868.19635020389202
25L2-Fabric Write BW4.968599175317518Gb/s1228.80.40434563601216783
26L2-Fabric Read Latency449.8986946067636Cycles
27L2-Fabric Write Latency144.17015133113895Cycles
28Wave Occupancy3303.5573155153597Wavefronts480068.82411073990332
29Instr Fetch BW683.181314670256Gb/s2307.07229.612483471268167
30Instr Fetch Latency18.659688472420168Cycles