JoseSantosAMD
da506ad9b5
* Updated links in documentation. (#328) Updated to reflect new GitHub organization. Fixed broken links to GitHub pages. Signed-off-by: David Galiffi <David.Galiffi@amd.com> * update branch for 2.x documentation builds Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * update checkout action and use concurrency instead of cancel-workflow-action Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * test addition of user option for container launch Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * remove --user option for container, try chown instead Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * fixing yaml syntax Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * reorder job step - start with checkout Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * restore missing run directive Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> * Update workloads to include log.txt Add missing MI200 workloads Signed-off-by: Jose Santos <josantos@amd.com> * Signed-off-by: Jose Santos <josantos@amd.com> Add vcopy workload for tests * Change exit codes for caught failures Signed-off-by: Jose Santos <josantos@amd.com> * reformat Signed-off-by: Jose Santos <josantos@amd.com> * Add pytest-xdist for pytest -n Signed-off-by: Jose Santos <josantos@amd.com> --------- Signed-off-by: David Galiffi <David.Galiffi@amd.com> Signed-off-by: Karl W. Schulz <karl.schulz@amd.com> Signed-off-by: Jose Santos <josantos@amd.com> Co-authored-by: David Galiffi <David.Galiffi@amd.com> Co-authored-by: Karl W. Schulz <karl.schulz@amd.com>
133 lines
7.7 KiB
Plaintext
133 lines
7.7 KiB
Plaintext
[roofline] Generating pmc_perf.csv (roofline counters only).
|
|
Omniperf version: 2.0.0-RC1
|
|
Profiler choice: rocprofv1
|
|
Path: /home1/josantos/omniperf/tests/workloads/kernel_names/MI200
|
|
Target: MI200
|
|
Command: ./tests/vcopy -n 1048576 -b 256 -i 3
|
|
Kernel Selection: None
|
|
Dispatch Selection: None
|
|
IP Blocks: All
|
|
|
|
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|
|
Collecting Performance Counters
|
|
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
[profiling] Current input file: tests/workloads/kernel_names/MI200/perfmon/pmc_perf_0.txt
|
|
|-> [rocprof] RPL: on '240321_162548' from '/opt/rocm-6.0.2' in '/home1/josantos/omniperf'
|
|
|-> [rocprof] RPL: profiling '""./tests/vcopy -n 1048576 -b 256 -i 3""'
|
|
|-> [rocprof] RPL: input file 'tests/workloads/kernel_names/MI200/perfmon/pmc_perf_0.txt'
|
|
|-> [rocprof] RPL: output dir '/tmp/rpl_data_240321_162548_4134506'
|
|
|-> [rocprof] RPL: result dir '/tmp/rpl_data_240321_162548_4134506/input0_results_240321_162548'
|
|
|-> [rocprof] ROCProfiler: input from "/tmp/rpl_data_240321_162548_4134506/input0.xml"
|
|
|-> [rocprof] gpu_index =
|
|
|-> [rocprof] kernel =
|
|
|-> [rocprof] range =
|
|
|-> [rocprof] 16 metrics
|
|
|-> [rocprof] SQ_INSTS_VALU_ADD_F16, SQ_INSTS_VALU_MUL_F16, SQ_INSTS_VALU_FMA_F16, SQ_INSTS_VALU_TRANS_F16, SQ_INSTS_VALU_ADD_F32, SQ_INSTS_VALU_MUL_F32, SQ_INSTS_VALU_FMA_F32, SQ_INSTS_VALU_TRANS_F32, TCP_TCC_READ_REQ_sum, TCP_TOTAL_CACHE_ACCESSES_sum, TCP_TCC_WRITE_REQ_sum, TCP_TCC_ATOMIC_WITH_RET_REQ_sum, TCC_EA_RDREQ_32B_sum, TCC_EA_RDREQ_sum, TCC_EA_WRREQ_64B_sum, TCC_EA_WRREQ_sum
|
|
|-> [rocprof] vcopy testing on GCD 0
|
|
|-> [rocprof] Finished allocating vectors on the CPU
|
|
|-> [rocprof] Finished allocating vectors on the GPU
|
|
|-> [rocprof] Finished copying vectors to the GPU
|
|
|-> [rocprof] sw thinks it moved 1.000000 KB per wave
|
|
|-> [rocprof] Total threads: 1048576, Grid Size: 4096 block Size:256, Wavefronts:16384:
|
|
|-> [rocprof] Launching the kernel on the GPU
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished copying the output vector from the GPU to the CPU
|
|
|-> [rocprof] Releasing GPU memory
|
|
|-> [rocprof] Releasing CPU memory
|
|
|-> [rocprof]
|
|
|-> [rocprof] ROCPRofiler: 3 contexts collected, output directory /tmp/rpl_data_240321_162548_4134506/input0_results_240321_162548
|
|
|-> [rocprof] File 'tests/workloads/kernel_names/MI200/pmc_perf_0.csv' is generating
|
|
|-> [rocprof]
|
|
[profiling] Current input file: tests/workloads/kernel_names/MI200/perfmon/pmc_perf_1.txt
|
|
|-> [rocprof] RPL: on '240321_162548' from '/opt/rocm-6.0.2' in '/home1/josantos/omniperf'
|
|
|-> [rocprof] RPL: profiling '""./tests/vcopy -n 1048576 -b 256 -i 3""'
|
|
|-> [rocprof] RPL: input file 'tests/workloads/kernel_names/MI200/perfmon/pmc_perf_1.txt'
|
|
|-> [rocprof] RPL: output dir '/tmp/rpl_data_240321_162548_4134693'
|
|
|-> [rocprof] RPL: result dir '/tmp/rpl_data_240321_162548_4134693/input0_results_240321_162548'
|
|
|-> [rocprof] ROCProfiler: input from "/tmp/rpl_data_240321_162548_4134693/input0.xml"
|
|
|-> [rocprof] gpu_index =
|
|
|-> [rocprof] kernel =
|
|
|-> [rocprof] range =
|
|
|-> [rocprof] 9 metrics
|
|
|-> [rocprof] SQ_INSTS_VALU_ADD_F64, SQ_INSTS_VALU_MUL_F64, SQ_INSTS_VALU_FMA_F64, SQ_INSTS_VALU_TRANS_F64, SQ_INSTS_VALU_MFMA_MOPS_F16, SQ_INSTS_VALU_MFMA_MOPS_BF16, SQ_INSTS_VALU_MFMA_MOPS_F32, SQ_INSTS_VALU_MFMA_MOPS_F64, TCP_TCC_ATOMIC_WITHOUT_RET_REQ_sum
|
|
|-> [rocprof] vcopy testing on GCD 0
|
|
|-> [rocprof] Finished allocating vectors on the CPU
|
|
|-> [rocprof] Finished allocating vectors on the GPU
|
|
|-> [rocprof] Finished copying vectors to the GPU
|
|
|-> [rocprof] sw thinks it moved 1.000000 KB per wave
|
|
|-> [rocprof] Total threads: 1048576, Grid Size: 4096 block Size:256, Wavefronts:16384:
|
|
|-> [rocprof] Launching the kernel on the GPU
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished copying the output vector from the GPU to the CPU
|
|
|-> [rocprof] Releasing GPU memory
|
|
|-> [rocprof] Releasing CPU memory
|
|
|-> [rocprof]
|
|
|-> [rocprof] ROCPRofiler: 3 contexts collected, output directory /tmp/rpl_data_240321_162548_4134693/input0_results_240321_162548
|
|
|-> [rocprof] File 'tests/workloads/kernel_names/MI200/pmc_perf_1.csv' is generating
|
|
|-> [rocprof]
|
|
[profiling] Current input file: tests/workloads/kernel_names/MI200/perfmon/pmc_perf_2.txt
|
|
|-> [rocprof] RPL: on '240321_162549' from '/opt/rocm-6.0.2' in '/home1/josantos/omniperf'
|
|
|-> [rocprof] RPL: profiling '""./tests/vcopy -n 1048576 -b 256 -i 3""'
|
|
|-> [rocprof] RPL: input file 'tests/workloads/kernel_names/MI200/perfmon/pmc_perf_2.txt'
|
|
|-> [rocprof] RPL: output dir '/tmp/rpl_data_240321_162549_4134881'
|
|
|-> [rocprof] RPL: result dir '/tmp/rpl_data_240321_162549_4134881/input0_results_240321_162549'
|
|
|-> [rocprof] ROCProfiler: input from "/tmp/rpl_data_240321_162549_4134881/input0.xml"
|
|
|-> [rocprof] gpu_index =
|
|
|-> [rocprof] kernel =
|
|
|-> [rocprof] range =
|
|
|-> [rocprof] 2 metrics
|
|
|-> [rocprof] SQ_LDS_IDX_ACTIVE, SQ_LDS_BANK_CONFLICT
|
|
|-> [rocprof] vcopy testing on GCD 0
|
|
|-> [rocprof] Finished allocating vectors on the CPU
|
|
|-> [rocprof] Finished allocating vectors on the GPU
|
|
|-> [rocprof] Finished copying vectors to the GPU
|
|
|-> [rocprof] sw thinks it moved 1.000000 KB per wave
|
|
|-> [rocprof] Total threads: 1048576, Grid Size: 4096 block Size:256, Wavefronts:16384:
|
|
|-> [rocprof] Launching the kernel on the GPU
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished copying the output vector from the GPU to the CPU
|
|
|-> [rocprof] Releasing GPU memory
|
|
|-> [rocprof] Releasing CPU memory
|
|
|-> [rocprof]
|
|
|-> [rocprof] ROCPRofiler: 3 contexts collected, output directory /tmp/rpl_data_240321_162549_4134881/input0_results_240321_162549
|
|
|-> [rocprof] File 'tests/workloads/kernel_names/MI200/pmc_perf_2.csv' is generating
|
|
|-> [rocprof]
|
|
[profiling] Current input file: tests/workloads/kernel_names/MI200/perfmon/timestamps.txt
|
|
|-> [rocprof] RPL: on '240321_162549' from '/opt/rocm-6.0.2' in '/home1/josantos/omniperf'
|
|
|-> [rocprof] RPL: profiling '""./tests/vcopy -n 1048576 -b 256 -i 3""'
|
|
|-> [rocprof] RPL: input file 'tests/workloads/kernel_names/MI200/perfmon/timestamps.txt'
|
|
|-> [rocprof] RPL: output dir '/tmp/rpl_data_240321_162549_4135068'
|
|
|-> [rocprof] RPL: result dir '/tmp/rpl_data_240321_162549_4135068/input0_results_240321_162549'
|
|
|-> [rocprof] ROCProfiler: input from "/tmp/rpl_data_240321_162549_4135068/input0.xml"
|
|
|-> [rocprof] gpu_index =
|
|
|-> [rocprof] kernel =
|
|
|-> [rocprof] range =
|
|
|-> [rocprof] 0 metrics
|
|
|-> [rocprof] vcopy testing on GCD 0
|
|
|-> [rocprof] Finished allocating vectors on the CPU
|
|
|-> [rocprof] Finished allocating vectors on the GPU
|
|
|-> [rocprof] Finished copying vectors to the GPU
|
|
|-> [rocprof] sw thinks it moved 1.000000 KB per wave
|
|
|-> [rocprof] Total threads: 1048576, Grid Size: 4096 block Size:256, Wavefronts:16384:
|
|
|-> [rocprof] Launching the kernel on the GPU
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished executing kernel
|
|
|-> [rocprof] Finished copying the output vector from the GPU to the CPU
|
|
|-> [rocprof] Releasing GPU memory
|
|
|-> [rocprof] Releasing CPU memory
|
|
|-> [rocprof]
|
|
|-> [rocprof] ROCPRofiler: 3 contexts collected, output directory /tmp/rpl_data_240321_162549_4135068/input0_results_240321_162549
|
|
|-> [rocprof] File 'tests/workloads/kernel_names/MI200/timestamps.csv' is generating
|
|
|-> [rocprof]
|
|
[roofline] Checking for roofline.csv in tests/workloads/kernel_names/MI200
|
|
[roofline] No roofline data found. Generating...
|
|
[roofline] Empirical Roofline PDFs saved!
|