search

ROCm / rocprofiler

ROC profiler library. Profiling with perf-counters and derived metrics.
https://rocm.docs.amd.com/projects/rocprofiler/en/latest/
MIT License
132 stars 49 forks source link

[Feature]: Ignore unsupported architectures #138

Open benrichard-amd opened 9 months ago

benrichard-amd commented 9 months ago

Suggestion Description

My laptop is an HP Omen. It has a gfx90c (not supported) and a gfx1032 (supported). The gfx90c cannot be disabled, even in the BIOS. I cannot run rocprof because it detects the gfx90c then fails:

terminate called after throwing an instance of 'rocprofiler::util::exception'
  what():  Translate(), ImportMetrics: bad block name 'GRBM', GFXIP is not supported(gfx90c)
/usr/local/bin/rocprofv2: line 304:  4006 Aborted                 (core dumped) LD_PRELOAD=$LD_PRELOAD:$ROCM_DIR/lib/rocprofiler/librocprofiler_tool.so $*

Shouldn't we be ignoring unsupported GPUs and only failing if a supported one cannot be detected?

Operating System

No response

GPU

No response

ROCm Component

No response

oldxie commented 5 months ago

any update ? rocminfo show that 2 devices (gfx1030 gfx90c) and I try to set ROCR_VISIBLE_DEVICES only gfx1030 but doesn't work

export ROCR_VISIBLE_DEVICES="1,GPU-****" && sudo rocprofv2 --list-counters terminate called after throwing an instance of 'rocprofiler::util::exception' what(): Translate(), ImportMetrics: bad block name 'GRBM', GFXIP is not supported(gfx90c) /usr/bin/rocprofv2: line 87: 828160 Aborted (core dumped) /opt/rocm-6.1.1/libexec/rocprofiler/ctrl