Files
rocm-systems/tests/workloads/ipblocks_SQ_SPI/MI200/pmc_perf.csv
T
Jose Santos 35ef8c0707 update workloads]
Signed-off-by: Jose Santos <josantos@amd.com>
2024-03-07 15:26:40 -06:00

5.1 KiB

1Dispatch_IDKernel_NameGPU_IDGrid_SizeWorkgroup_SizeLDS_Per_WorkgroupScratch_Per_WorkitemArch_VGPRAccum_VGPRSGPRwave_sizeobjSQ_ACTIVE_INST_ANYSQ_CYCLESSQ_BUSY_CYCLESSQ_BUSY_CU_CYCLESSQ_ACTIVE_INST_VMEMSQ_ACTIVE_INST_LDSSQ_ACTIVE_INST_VALUSQ_ACTIVE_INST_SCASPI_RA_WAVE_SIMD_FULL_CSNSPI_RA_VGPR_SIMD_FULL_CSNwave_size_1obj_1SQ_INSTS_LDSSQ_INSTS_GDSSQ_INSTS_EXP_GDSSQ_INSTS_BRANCHSQ_INSTS_SENDMSGSQ_WAVE_CYCLESSQ_WAIT_ANYSQ_WAIT_INST_ANYSPI_RA_RES_STALL_CSNSPI_RA_TMP_STALL_CSNwave_size_2obj_2SQ_INSTS_VALU_ADD_F64SQ_INSTS_VALU_MUL_F64SQ_INSTS_VALU_FMA_F64SQ_INSTS_VALU_TRANS_F64SQ_INSTS_VALU_INT32SQ_INSTS_VALU_INT64SQ_INSTS_SMEMSQ_INSTS_FLATSPI_RA_REQ_NO_ALLOCSPI_RA_REQ_NO_ALLOC_CSNwave_size_3obj_3SQC_ICACHE_MISSES_DUPLICATESQC_DCACHE_INPUT_VALID_READYBSQC_DCACHE_ATOMICSQC_DCACHE_REQ_READ_8SQC_DCACHE_REQSQC_DCACHE_HITSSQC_DCACHE_MISSESSQC_DCACHE_MISSES_DUPLICATEwave_size_4obj_4SQ_INSTS_VALU_ADD_F16SQ_INSTS_VALU_MUL_F16SQ_INSTS_VALU_FMA_F16SQ_INSTS_VALU_TRANS_F16SQ_INSTS_VALU_ADD_F32SQ_INSTS_VALU_MUL_F32SQ_INSTS_VALU_FMA_F32SQ_INSTS_VALU_TRANS_F32GRBM_SPI_BUSYSPI_CSN_NUM_THREADGROUPSSPI_CSN_WAVEwave_size_5obj_5SQ_INSTS_VALU_CVTSQ_INSTS_VMEM_WRSQ_INSTS_VMEM_RDSQ_INSTS_VMEMSQ_INSTS_SALUSQ_INSTS_VSKIPPEDSQ_INSTSSQ_INSTS_VALUGRBM_COUNTGRBM_GUI_ACTIVESPI_CSN_WINDOW_VALIDSPI_CSN_BUSYwave_size_6obj_6SQC_DCACHE_REQ_READ_1SQC_DCACHE_REQ_READ_2SQC_DCACHE_REQ_READ_4wave_size_7obj_7SQ_IFETCHSQ_LDS_BANK_CONFLICTSQ_LDS_ADDR_CONFLICTSQ_LDS_UNALIGNED_STALLSQ_WAVESSQ_WAVES_EQ_64SQ_WAVES_LT_64SQ_WAVES_LT_48SPI_RA_BAR_CU_FULL_CSNSPI_RA_TGLIM_CU_FULL_CSNwave_size_8obj_8SQ_INSTS_SMEM_NORMSQ_INSTS_MFMASQ_INSTS_VALU_MFMA_I8SQ_INSTS_VALU_MFMA_F16SQ_INSTS_VALU_MFMA_BF16SQ_INSTS_VALU_MFMA_F32SQ_INSTS_VALU_MFMA_F64SQ_VALU_MFMA_BUSY_CYCLESSPI_VWC_CSC_WRSPI_RA_BULKY_CU_FULL_CSNwave_size_9obj_9SQC_TC_DATA_WRITE_REQSQC_TC_DATA_ATOMIC_REQSQC_TC_STALLSQC_TC_REQSQC_DCACHE_REQ_READ_16SQC_ICACHE_REQSQC_ICACHE_HITSSQC_ICACHE_MISSESwave_size_10obj_10SQ_INSTS_FLAT_LDS_ONLYSQ_INSTS_VALU_MFMA_MOPS_I8SQ_INSTS_VALU_MFMA_MOPS_F16SQ_INSTS_VALU_MFMA_MOPS_BF16SQ_INSTS_VALU_MFMA_MOPS_F32SQ_INSTS_VALU_MFMA_MOPS_F64SQC_TC_INST_REQSQC_TC_DATA_READ_REQwave_size_11obj_11SQ_WAVES_LT_32SQ_WAVES_LT_16SQ_ITEMSSQ_LDS_MEM_VIOLATIONSSQ_LDS_ATOMIC_RETURNSQ_LDS_IDX_ACTIVESQ_WAVES_RESTOREDSQ_WAVES_SAVEDSPI_RA_WVLIM_STALL_CSNSPI_SWC_CSC_WRwave_size_12obj_12SQ_ACTIVE_INST_EXP_GDSSQ_ACTIVE_INST_MISCSQ_ACTIVE_INST_FLATSQ_INST_CYCLES_VMEM_WRSQ_INST_CYCLES_VMEM_RDSQ_INST_CYCLES_SMEMSQ_INST_CYCLES_SALUSQ_THREAD_CYCLES_VALUSPI_RA_SGPR_SIMD_FULL_CSNSPI_RA_LDS_CU_FULL_CSNStart_TimestampEnd_Timestamp
20vecCopy(double*, double*, double*, int, int) 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
31vecCopy(double*, double*, double*, int, int) 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
42vecCopy(double*, double*, double*, int, int) 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