Skip to content

[CI] Refresh parity trunk job names - #3650

Merged
jithunnair-amd merged 1 commit into
developfrom
ethanwee1-fix-parity-canonical-run
Sep 16, 2026
Merged

jithunnair-amd merged 1 commit into
developfrom
ethanwee1-fix-parity-canonical-run

Conversation

@ethanwee1

Copy link
Copy Markdown

Problem

The MI350 parity auto-run for a3fcc26e failed because parity_job_config.json still identified trunk as CUDA 13.0 and Jammy ROCm Python 3.10. download_testlogs therefore rejected the current trunk jobs, selected a scheduled periodic run carrying the stale CUDA name, and found no matching default or distributed artifacts.

Tracked in AIPYTORCH-1223.

Changes

  • Update trunk CUDA default/distributed matching and the auto-trigger regex from CUDA 13.0 to CUDA 13.2.
  • Update MI350 trunk default, distributed fallback, and Inductor matching to Noble ROCm Python 3.11.
  • Leave the standalone rocm-mi350 and periodic-rocm-mi350 Python 3.12 sources unchanged.

Validation

  • Parsed parity_job_config.json and verified the new CUDA regex accepts CUDA 13.2 while rejecting the stale CUDA 13.0 identity.
  • Re-ran the original SHA on MI350 from this branch: https://github.com/ROCm/pytorch/actions/runs/35132214717
  • The run selected upstream trunk run 35098139093, found 25 CUDA test jobs, downloaded all ROCm and CUDA default/distributed/Inductor artifacts, generated the CSV, and uploaded the summary successfully.

Match the current CUDA 13.2 and Noble ROCm Python 3.11 trunk jobs so parity selects the full push run instead of a scheduled periodic run.
],
"inductor": [
{ "workflow": "trunk", "job_prefix": "linux-jammy-rocm-py3.10-mi350" }
{ "workflow": "trunk", "job_prefix": "linux-noble-rocm-py3.11-mi350" }

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

thinking we could make all this more robust to account for future changes/release version names?

@jithunnair-amd
jithunnair-amd marked this pull request as ready for review September 16, 2026 23:08
@jithunnair-amd
jithunnair-amd merged commit 6779019 into develop Sep 16, 2026
7 checks passed
@jithunnair-amd
jithunnair-amd deleted the ethanwee1-fix-parity-canonical-run branch September 16, 2026 23:08
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