-
Couldn't load subscription status.
- Fork 74
[Liger-Kernels] Fix Liger-Kernels tests hanging #5382
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
|
Since my patch got merged upstream now this PR only does refactoring |
| echo "****** Running Liger-Kernel tests ******" | ||
| echo "************************************************" | ||
|
|
||
| run_pytest_command -vvv -n ${PYTEST_MAX_PROCESSES:-4} Liger-Kernel/test/ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Liger hangs with -n 4 for some reason, but passes without parallelization after merging cache cleanup
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So I make tests sequential
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In my local experiment that doesn't change the total runtime, maybe due to the GPU being the bottleneck. We do have 1 GPU for these tests
|
CI pass without hanging: https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/18785052372 |
Should we update the PR title? |
This PR still changes number of runners ( |
Resolves #5292
There's also refactoring of Liger-Kernels installation to be consistent with sglang, vllm
PR depends on:
Before merging: