[GPU] Implement CDNA block intrinsics (4/6)#23977
Merged
nirvedhmeshram merged 1 commit intoiree-org:mainfrom Apr 1, 2026
Merged
[GPU] Implement CDNA block intrinsics (4/6)#23977nirvedhmeshram merged 1 commit intoiree-org:mainfrom
nirvedhmeshram merged 1 commit intoiree-org:mainfrom
Conversation
Block intrinsics with a single-element accumulator (e.g. 4x4 with 16 blocks) require the acc to be extracted to a scalar before passing to amdgpu.mfma, and the scalar result broadcast back to the vector type. Part of iree-org#23941 Signed-off-by: Nirvedh Meshram <nirvedh@gmail.com> Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
nirvedhmeshram
requested review from
Groverkss,
Max191,
krzysz00 and
qedawkins
as code owners
6 tasks
qedawkins
approved these changes
Apr 1, 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Intrinsics with a single-element accumulator (e.g. 4x4 f64 with 4 blocks) require the acc to be extracted to a scalar before passing to amdgpu.mfma, and the scalar result broadcast back to the vector type. This is because otherwise we don't have a valid result type as per the op definition.
Part of #23941