Skip to content

[ROCm][CI] Route unittests to MI350 ecosystem runners - #9567

Closed
amdfaa wants to merge 1 commit into
pytorch:mainfrom
amdfaa:rocm-ci-amd-vtr-gfx942
Closed

[ROCm][CI] Route unittests to MI350 ecosystem runners#9567
amdfaa wants to merge 1 commit into
pytorch:mainfrom
amdfaa:rocm-ci-amd-vtr-gfx942

Conversation

@amdfaa

@amdfaa amdfaa commented Jul 27, 2026

Copy link
Copy Markdown
Contributor

Summary

Switch the ROCm unittests-rocm job in .github/workflows/tests.yml to the
shared MI350 (gfx950) ecosystem runner scale set used by downstream PyTorch
ecosystem projects (e.g. pytorch/helion, pytorch/torchtitan):

  • runner label -> linux.rocm.gpu.ecosystem.gfx950.1

Test plan

  • ROCm unittests-rocm job should schedule and run on a
    linux.rocm.gpu.ecosystem.gfx950.1 runner (same-repo PR / push /
    workflow_dispatch; the job is skipped for fork PRs by design because fork PRs
    don't receive a GitHub OIDC id-token for ECR/AWS auth).

Made with Cursor

cc @jeffdaily @jithunnair-amd

@pytorch-bot

pytorch-bot Bot commented Jul 27, 2026

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/9567

Note: Links to docs will display an error until the docs builds have been completed.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@pytorch-bot

pytorch-bot Bot commented Jul 27, 2026

Copy link
Copy Markdown

The following ciflow label(s) have been added but CI has not been triggered yet because the workflows are awaiting approval:

  • ciflow/rocm

Once a maintainer approves the workflows (scroll to the bottom of the PR page), the corresponding CI jobs will be triggered automatically. Please ping one of the reviewers if you do not have access to approve and run workflows.

@amdfaa amdfaa changed the title [ROCm][CI] Route gfx942 unittests to amd-vtr (Vultr) runners [ROCm][CI] Route unittests to rocOps ecosystem runners Jul 27, 2026
@amdfaa amdfaa changed the title [ROCm][CI] Route gfx942 unittests to amd-vtr (Vultr) runners [ROCm][CI] Route unittests to rocOps ecosystem runners Jul 27, 2026
Switch the unittests-rocm job in .github/workflows/tests.yml to the
MI350 (gfx950) ecosystem runner label linux.rocm.gpu.ecosystem.gfx950.1,
used by downstream PyTorch ecosystem projects.

Made with Cursor
@amdfaa
amdfaa force-pushed the rocm-ci-amd-vtr-gfx942 branch from f41e543 to 70575a6 Compare July 27, 2026 19:24
@amdfaa amdfaa changed the title [ROCm][CI] Route unittests to rocOps ecosystem runners [ROCm][CI] Route unittests to MI350 ecosystem runners Jul 27, 2026
branches:
- nightly
- main
- release/*

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Suggested change
- release/*
tags:
- ciflow/rocm/*

Let's also add the tag push trigger so PRs can run ROCm CI

@NicolasHug

Copy link
Copy Markdown
Member

Thanks for the PR - I won't merge it because I'm in the process of moving the ROCJPEG decoder into TorchCodec (and I'll remove it from TorchVision) in meta-pytorch/torchcodec#1554. Since this job is just testing the ROCJPEG decoder, it'll soon be removed.

But I'll follow-up in TorchCodec with the runner update. Thanks!

@NicolasHug NicolasHug closed this Jul 28, 2026
@jithunnair-amd

Copy link
Copy Markdown
Contributor

@NicolasHug Thanks for the update

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.

3 participants