GPU count includes amd-smi warning lines, inflating MAD_SYSTEM_NGPUS
Open
Beginner friendly
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 9
- Forks
- 18
- Avg merge
- 5d 8h
- Merged PRs (30d)
- 8
Description
rocm_tool_manager.py:281 counts GPUs with amd-smi list --csv | tail -n +3 | wc -l. Without the render/video groups amd-smi prints one WARNING per GPU on stdout, and those lines get counted:
WARNING: User is missing the following required groups: render, video. (x9)
$ amd-smi list --csv | tail -n +3 | wc -l # node has 8 GPUs
17
Fails tests/integration/test_gpu_management.py:492 (test_renderD_count_matches_gpu_count): assert 8 == 17.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start at rocm_tool_manager.py:281 and read the GPU-counting command, then run tests/integration/test_gpu_management.py:492 to reproduce the mismatch. Confirm the resulting GPU count matches the render device count when amd-smi emits warning lines, while the existing behavior remains correct in the normal case.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- cli
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Active
- Clarity
- Clearly specified
- Newbie friendliness
- 88/100