ROCm / rocprofiler

ROC profiler library. Profiling with perf-counters and derived metrics.
https://rocm.docs.amd.com/projects/rocprofiler/en/latest/
MIT License
126 stars 46 forks source link

ROCPRofiler: 0 contexts collected #61

Closed Madu86 closed 2 years ago

Madu86 commented 2 years ago

hello, I am new to rocprof and trying to profile MatrixTranspose example (https://github.com/ROCm-Developer-Tools/HIP/blob/develop/samples/2_Cookbook/0_MatrixTranspose/MatrixTranspose.cpp). My input file is the same from help menu with the exception of kernel name, however, it seems that rocprof is not collecting any metrics for some reason. I tried with different examples but faced the same issue. Does this mean that I dont have access to performance counters or something? Thanks!

 ROCProfiler: input from "/tmp/rpl_data_211123_102928_104567/input0.xml" gpu_index = 0,1,2,3 kernel = matrixTranspose range = 1:4 10 metrics Wavefronts, VALUInsts, SALUInsts, SFetchInsts, FlatVMemInsts, LDSInsts, FlatLDSInsts, GDSInsts, VALUUtilization, FetchSize Device name PASSED!

ROCPRofiler: 0 contexts collected, output directory /tmp/rpl_data_211123_102928_104567/input0_results_211123_102928 RPL: result dir '/tmp/rpl_data_211123_102928_104567/input1_results_211123_102928' ROCProfiler: input from "/tmp/rpl_data_211123_102928_104567/input1.xml" gpu_index = 0,1,2,3 kernel = matrixTranspose range = 1:4 2 metrics WriteSize, L2CacheHit Device name PASSED!

ROCPRofiler: 0 contexts collected, output directory /tmp/rpl_data_211123_102928_104567/input1_results_211123_102928

Madu86 commented 2 years ago

Found what went wrong!