Skip to content

Conversation

@jakub-sochacki
Copy link
Contributor

  • Add linux.hpu.gaudi3.8 runner to benchmark matrix (placeholder)
  • Create test-[throughput | latency | serving]-hpu.json files with benchmark configurations
  • Add gaudi3 to default runners list in workflow dispatch
  • Implement HPU device detection using hl-smi command
  • Add HPU-specific Docker image suffix (-hpu)
  • Add special commit selection for HPU that queries vllm-gaudi history
  • Clone vllm-gaudi vllm/last-good-commit-for-vllm-gaudi branch to get compatibility history (N most recent VLLM_STABLE_COMMIT updates)
  • Search through commit history to find most recent vLLM version with existing Docker image built by CI

The commit selection mechanism solves the race condition where VLLM_STABLE_COMMIT might change between CI image builds and benchmark runs (every 12 hours), ensuring benchmarks always find an existing compatible Docker image.

@meta-cla
Copy link

meta-cla bot commented Oct 15, 2025

Hi @jakub-sochacki!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!

@meta-cla meta-cla bot added the cla signed label Oct 15, 2025
@meta-cla
Copy link

meta-cla bot commented Oct 15, 2025

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!

Copy link
Collaborator

@louie-tsai louie-tsai left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

looks good to me. thanks

pip install -r .github/scripts/requirements.txt \
--extra-index-url https://download.pytorch.org/whl/rocm6.3
elif [[ "${DEVICE_NAME}" == "hpu" ]]; then
grep -v "^torch==" .github/scripts/requirements.txt > /tmp/requirements_no_torch.txt
Copy link
Contributor

@huydhn huydhn Oct 28, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nit: Is there a way to write something as follows in the requirements file?

torch; platform_system == 'Linux' and platform_machine != 'hpu'

Probably no, I haven't seen this syntax before, so just want to check

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think we need to install pytorch here now, but let me follow up on this in a separate PR to just remove torch from requirements.txt

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Alright @huydhn, so please let me know if you removed torch from requirements there in other PR. Then i will remove the line which filters out torch for gaudi.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@huydhn , this change is to make sure we don't re-install torch. Since HPU is using special torch version. I hope it is OK to leave the grep -v "^torch==" .github/scripts/requirements.txt > /tmp/requirements_no_torch.txt just in case.

Copy link
Contributor

@huydhn huydhn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! Plz rebase and I can help land this

@huydhn huydhn merged commit 55c19ce into pytorch:main Oct 31, 2025
2 of 30 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants