coleramos425
230 righe
12 KiB
Plaintext
230 righe
12 KiB
Plaintext
Omniperf version: 2.0.0
|
|
Profiler choice: rocprofv2
|
|
Path: /home/colramos/omniperf/tests/workloads/kernel/MI300X_A1
|
|
Target: MI300X_A1
|
|
Command: ./tests/vcopy -n 1048576 -b 256 -i 3
|
|
Kernel Selection: ['"vecCopy(double*,', 'double*,', 'double*,', 'int,', 'int)', '[clone', '.kd]"']
|
|
Dispatch Selection: None
|
|
Hardware Blocks: All
|
|
|
|
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|
|
Collecting Performance Counters
|
|
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/SQ_IFETCH_LEVEL.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - GRBM_COUNT
|
|
|-> [/opt/rocm/bin/rocprofv2] - GRBM_GUI_ACTIVE
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_IFETCH
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_IFETCH_LEVEL
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/SQ_INST_LEVEL_LDS.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_LDS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_LEVEL_LDS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACCUM_PREV_HIRES
|
|
|-> [/opt/rocm/bin/rocprofv2] Enabling Counter Collection
|
|
|-> [/opt/rocm/bin/rocprofv2] vcopy testing on GCD 0
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/SQ_INST_LEVEL_SMEM.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_SMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_LEVEL_SMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACCUM_PREV_HIRES
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/SQ_INST_LEVEL_VMEM.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_LEVEL_VMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACCUM_PREV_HIRES
|
|
|-> [/opt/rocm/bin/rocprofv2] Enabling Counter Collection
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/SQ_LEVEL_WAVES.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - GRBM_COUNT
|
|
|-> [/opt/rocm/bin/rocprofv2] - GRBM_GUI_ACTIVE
|
|
|-> [/opt/rocm/bin/rocprofv2] - CPC_ME1_BUSY_FOR_PACKET_DECODE
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_CYCLES
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_0.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_CYCLES
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_BUSY_CYCLES
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_BUSY_CU_CYCLES
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_1.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_SALU
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VSKIPPED
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_10.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_TC_DATA_ATOMIC_REQ
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_TC_STALL
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_TC_REQ
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_REQ_READ_16
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_11.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_INPUT_VALID_READYB
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_ATOMIC
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_REQ_READ_8
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_REQ
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_HITS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_MISSES
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_12.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_REQ_READ_2
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_DCACHE_REQ_READ_4
|
|
|-> [/opt/rocm/bin/rocprofv2] Enabling Counter Collection
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_13.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_ATOMIC[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_BUBBLE[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_CYCLE[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_ATOMIC[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_ATOMIC[1]
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_14.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_ATOMIC_LEVEL[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_RDREQ[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_RDREQ_32B[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_RDREQ_LEVEL[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_ATOMIC_LEVEL[1]
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_15.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_WRREQ[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_WRREQ_64B[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_WRREQ_LEVEL[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_HIT[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_EA0_WRREQ[1]
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_16.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_MISS[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_READ[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_REQ[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_RW_REQ[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_MISS[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_READ[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_REQ[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_RW_REQ[1]
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_17.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TAG_STALL[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TOO_MANY_EA_WRREQS_STALL[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_WRITE[0]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TAG_STALL[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TOO_MANY_EA_WRREQS_STALL[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_WRITE[1]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TAG_STALL[2]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TOO_MANY_EA_WRREQS_STALL[2]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_WRITE[2]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TAG_STALL[3]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TOO_MANY_EA_WRREQS_STALL[3]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_WRITE[3]
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCC_TAG_STALL[4]
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_2.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_TRANS_F16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_ADD_F32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MUL_F32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_FMA_F32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_TRANS_F32
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_3.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_TRANS_F64
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_INT32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_INT64
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_SMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_FLAT
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_4.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_BRANCH
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_SENDMSG
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAIT_ANY
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAIT_INST_ANY
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_ANY
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_VMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_LDS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_VALU
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCP_UTCL1_TRANSLATION_MISS_sum
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCP_UTCL1_TRANSLATION_HIT_sum
|
|
|-> [/opt/rocm/bin/rocprofv2] - TCP_UTCL1_PERMISSION_MISS_sum
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_5.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_SCA
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_EXP_GDS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_MISC
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ACTIVE_INST_FLAT
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_CYCLES_VMEM_WR
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_CYCLES_VMEM_RD
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_CYCLES_SMEM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INST_CYCLES_SALU
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_6.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_THREAD_CYCLES_VALU
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_IFETCH
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_BANK_CONFLICT
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_ADDR_CONFLICT
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_UNALIGNED_STALL
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES_EQ_64
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES_LT_64
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES_LT_48
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_7.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES_LT_32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_WAVES_LT_16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_ITEMS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_MEM_VIOLATIONS
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_ATOMIC_RETURN
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_LDS_IDX_ACTIVE
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_8.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_SMEM_NORM
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_MFMA
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_I8
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_F16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_BF16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_F32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_F64
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/pmc_perf_9.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] ROCProfilerV2: Collecting the following counters:
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_MOPS_I8
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_MOPS_F16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_MOPS_BF16
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_MOPS_F32
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQ_INSTS_VALU_MFMA_MOPS_F64
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_TC_INST_REQ
|
|
|-> [/opt/rocm/bin/rocprofv2] - SQC_TC_DATA_READ_REQ
|
|
|
|
[profiling] Current input file: tests/workloads/kernel/MI300X_A1/perfmon/timestamps.txt
|
|
|-> [/opt/rocm/bin/rocprofv2] vcopy testing on GCD 0
|
|
|-> [/opt/rocm/bin/rocprofv2] Finished allocating vectors on the CPU
|
|
|-> [/opt/rocm/bin/rocprofv2] Finished allocating vectors on the GPU
|
|
|-> [/opt/rocm/bin/rocprofv2] Finished copying vectors to the GPU
|
|
|-> [/opt/rocm/bin/rocprofv2] sw thinks it moved 1.000000 KB per wave
|
|
|-> [/opt/rocm/bin/rocprofv2] Total threads: 1048576, Grid Size: 4096 block Size:256, Wavefronts:16384:
|
|
|-> [/opt/rocm/bin/rocprofv2] Launching the kernel on the GPU
|
|
|-> [/opt/rocm/bin/rocprofv2] Finished executing kernel
|
|
[roofline] Roofline temporarily disabled in MI300
|