[Docs] Point the README quick-check at the GEMM correctness test#647
Merged
Conversation
The Run tests section attached a "~15s GEMM correctness" comment to `bash scripts/run_tests.sh`, but that script runs the entire suite (3000+ tests plus example scripts and MLIR FileCheck) and takes many minutes, so the label is misleading for a post-install sanity check. Point the line at the actual GEMM correctness test instead. It runs the real GPU MMA path in ~10s (`not large_shape`), matches the original intent of the comment, and avoids the full suite (including the 8-GPU allreduce tests). See ROCm#646.
a927e3c to
3800ee0
Compare
coderfeli
approved these changes
Jun 4, 2026
Collaborator
|
Thanks @Arist12 |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
The README "Run tests" section labels
bash scripts/run_tests.shas a quick post-install check:# Run GEMM correctness tests (fast, ~15s) bash scripts/run_tests.shBut
run_tests.shruns the entire test suite, which takes many minutes — the "~15s GEMM correctness" label is inaccurate and misleads contributors validating a fresh build. This PR points the line at the actual GEMM correctness test, which runs the real GPU path in ~10s and matches the original intent of the comment.Closes #646
Findings
run_tests.shruns the full suite, in three phases:pytestovertests/kernels|unit|system|python/examples(3118 tests selected with the default-m "not large_shape"), then all standaloneexamples/scripts, then every
tests/mlir/**/*.mlirFileCheck test.tests/kernels/test_allreduce.pyalone (8-GPU accuracy tests,@pytest.mark.multi_gpu) did not finish 6 parametrizations in 150s on an 8-GPU host. Those multi-GPU tests auto-skip on <4 GPUs but run (and dominate) on larger hosts.git blametraces the comment to the v0.1 PR (Pr/v0.1 #164), when the suite was small. The targeted GEMM correctness test still runs in ~10s today, so keeping the comment and swapping the command restores its accuracy.Change (one line)
Replace
bash scripts/run_tests.shwith the GEMM correctness test:Verification