Add more tests#537
Open
micmelesse wants to merge 2 commits into
Open
Conversation
micmelesse
force-pushed
the
micmelesse/ccl_capture_main
branch
from
bca0178 to
cc51e19
Compare
…l_gather & all_to_all
micmelesse
force-pushed
the
micmelesse/ccl_capture_main
branch
from
cc51e19 to
dda5cc2
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
Note
Copilot was unable to run its full agentic suite in this review.
Expands collective-communication test coverage to validate correctness across different implementations and execution modes, including repeated graph replays with varying inputs.
Changes:
- Refactors all-gather and all-to-all tests into a replay-based harness that checks per-rank/per-chunk correctness across many iterations.
- Adds parameterization over implementation (torch/triton/gluon), execution mode (eager w/ barrier, eager w/o barrier, graph), and whether inputs vary between replays.
- Introduces CUDA graph capture + replay paths for the Iris-based collectives (and torch in all-gather).
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 8 comments.
| File | Description |
|---|---|
| tests/ccl/test_all_to_all_gluon.py | Adds replay/graph coverage for all-to-all across triton vs gluon and varying inputs, with per-source-chunk validation. |
| tests/ccl/test_all_gather_gluon.py | Refactors all-gather into a shared replay harness, expands parameter matrix (incl. torch), and adds graph replay + varying-input coverage. |
micmelesse
force-pushed
the
micmelesse/ccl_capture_main
branch
from
9b4c800 to
a0360ec
Compare
mawad-amd
approved these changes
Jun 18, 2026
Contributor
Author
|
@mawad-amd The test failures are the cases that I added. @aamarnat wanted me to see if I could recreate the correctness issues that I was seeing on top of main. |
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
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.
Motivation
This pr expands test coverage from a single eager call to back to back calls with both eager and graph capture. We also add a flag for varying input. The additional tests fail locally. It seems there is a correctness issue with the kernels when running in non eager modes.
Technical Details
Test Plan
Test Result
Submission Checklist