[Test] Add/update GEMM tests to be consistent for performance.#1174
Merged
mshahneo merged 1 commit intoApr 29, 2026
Merged
Conversation
Updates the GEMM test cases to have: - Input range from `0.0 to 1.0` - Use large register file - Remove the `gpu.barrier` to sync threads, this barrier is only needed so that some threads do not go too far forward with the prefetching that other threads necessary prefetched data is thrown out of the cache. However, doing `gpu.barrier` is very expensive. Until we have something of a` named_barrier` type available, skipping this barrier proved to be more performant.
Garra1980
approved these changes
Apr 29, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Updates the GEMM test cases to have:
0.0 to 1.0gpu.barrierto sync threads, this barrier is only needed so that some threads do not go too far forward with the prefetching that other threads necessary prefetched data is thrown out of the cache. However, doinggpu.barrieris very expensive. Until we have something of anamed_barriertype available, skipping this barrier proved to be more performant.Please review these guidelines to help with the review process: