Skip to content

Backport: Mount /home/runner/_work in ROCm workflow containers - #841

Open
mkunredd wants to merge 4 commits into
rocm-jaxlib-v0.11.1-alphafrom
backport-workdir-volume-to-release
Open

mkunredd wants to merge 4 commits into
rocm-jaxlib-v0.11.1-alphafrom
backport-workdir-volume-to-release

Conversation

@mkunredd

Copy link
Copy Markdown

Summary

Backport of #840 to rocm-jaxlib-v0.11.1-alpha.

Adds /home/runner/_work:/home/runner/_work volume mounts to pytest_rocm.yml and bazel_rocm.yml so gfx950.8 container jobs can initialize without runner communication failures.

Validation

Test plan

  • Merge and re-run CI - Wheel Tests (Nightly/Release) on release branch
  • Confirm gfx950.8 Pytest ROCm passes Initialize containers and completes pytest

mminutoli and others added 4 commits August 12, 2026 18:04
Upstream CI has stopped publishing jax/jaxlib nightly wheels to GCS
(gs://jax-nightly-artifacts/latest), leaving the ROCm nightly test legs
running against missing/stale base wheels.

On the ROCm fork, build the base jax and jaxlib wheels on the same
amd-do-linux.jax.cpu runners as the ROCm plugin/pjrt wheels and upload
them to the same S3 prefix, so the ROCm test legs consume freshly built
wheels instead of the broken GCS bucket. Upstream (jax-ml) behavior is
unchanged.

- ci/build_rocm_artifacts.sh: allow "jax"/"jaxlib" artifacts; skip
  auditwheel for the pure-python jax wheel.
- wheel_tests_nightly_release.yml: new fork-only build-rocm-jax-jaxlib
  job; wire it into the ROCm test legs and select S3 as the jax/jaxlib
  source on the fork.
- pytest_rocm.yml / bazel_rocm.yml: new optional jax-jaxlib-source input
  (default gcs), forwarded to the download action.
- download-jax-rocm-wheels: when source is s3, fetch jax/jaxlib from the
  CloudFront listing alongside the ROCm wheels instead of GCS.
This adds 8028c6c78ea42b7184be31bcbf9d1fe40809bde1 on top of upstream's pin.
Ensure GitHub Actions workspace paths are available inside ROCm container jobs by mounting /home/runner/_work in pytest_rocm and bazel_rocm workflows.

Co-authored-by: Cursor <cursoragent@cursor.com>
@mminutoli
mminutoli force-pushed the rocm-jaxlib-v0.11.1-alpha branch 9 times, most recently from 4f9521c to a94e1bb Compare August 17, 2026 18:27
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants