sycl: Batched mulmat rework for oneDNN dispatch #14617
Merged
Conversation
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
Alcpz
reviewed
Jul 11, 2025
747c12e to
b2785f8
Compare
Alcpz
approved these changes
Jul 11, 2025
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Outdated suggestions cannot be applied.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.
PR proposes a rework to existing dispatch of batched Mul_Mats for the sycl backend to oneDNN to allow better use of broadcasts for non matching batch sizes for inputs and handle non continuous data being passed in the tensors. This reduces the number of calls to oneDNN matmul.
Additionaly small fix added to PR for the
ggml_sycl_mul_mat_vec_nccase to allowsrc1to be non-continuous as well.test-backend-opspassing all testsllama-benchrunning model qwen2 shows no performance regression compared to masterrunning on Intel Battlemage