Conversation
Use trunk as the sole MI350 source and derive each run's actual CUDA and ROCm job families so current and baseline commits survive job-name and shard-count changes.
Scope fallback checks to trunk push runs, require complete shard families, preserve discovered test kinds, and retain configured baseline fallbacks for other architectures.
| _TEST_JOB = re.compile( | ||
| r"^(?P<prefix>.+) / (?P<kind>test(?:-osdc)?) " | ||
| r"\((?P<config>[^,]+), (?P<shard>\d+), (?P<total>\d+)" | ||
| ) |
Collaborator
There was a problem hiding this comment.
Some examples in a comment would be helpful to parse these regexes
| r"^(?P<prefix>.+) / (?P<kind>test(?:-osdc)?) " | ||
| r"\((?P<config>[^,]+), (?P<shard>\d+), (?P<total>\d+)" | ||
| ) | ||
| _SPECIAL_FAMILIES = ("build-only", "debug", "no-ops", "slow-gradcheck", "smoke") |
Collaborator
There was a problem hiding this comment.
hardcoded much? does it pick up junk if we don't have this?
| f"(last HTTP {response.status_code}) for {jobs_url}" | ||
| ) | ||
|
|
||
| def get_workflow_jobs(wf, all_attempts=False): |
Collaborator
There was a problem hiding this comment.
should the default value for all_attempts be True instead? Why not?
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
MI350 default, distributed, and Inductor now run together in
trunk.yml; the standalonerocm-mi350andperiodic-rocm-mi350workflows are manual-only metrics workflows with automatic triggers disabled. Parity still models them as primary/fallback sources and hardcodes job prefixes and shard totals, so historical comparisons and upstream environment renames can select the wrong run or miss valid jobs.Tracked in AIPYTORCH-1223. Stacked on #3651; #3650 remains unchanged.
Changes
test/test-osdcinto log and artifact names.Validation
python -m unittest -v test_job_name_match.py(7 tests passed)python -m py_compile download_testlogs job_name_match.py test_job_name_match.pypython -m ruff check job_name_match.py test_job_name_match.pya3fcc26e): https://github.com/ethanwee1/pytorch/actions/runs/35143014867b285e55avs58acd33b): https://github.com/ethanwee1/pytorch/actions/runs/35143014987All four hardened-head workflows passed with source logs enabled. The current runs selected trunk
35098139093; the historical comparisons independently selected current trunk34612152506and baseline trunk34610386489, detected the older Jammy ROCm Python 3.10 family, and downloaded all baseline distributed shards. The personal-fork historical report contains 5,172 distributed and 369,710 total tests on each commit.