Enable autotuning and bf16 accumulation for SYCL CUTLASS #4
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
aacostadiaz
approved these changes
May 5, 2025
Enable autotuning for SYCL CUTLASS by completing the SYCL benchmark request class. Also adds a temporary workaround to allow bf16 GEMM to accumulate in FP32 in code paths used when auto-tuning is active. Signed-off-by: Lukas Sommer <[email protected]>
Signed-off-by: Lukas Sommer <[email protected]>
sommerlukas
force-pushed
the
cutlass-sycl-autotune
branch
from
e21c49d to
d76676d
Compare
sommerlukas
changed the title
|
This PR depends on codeplaysoftware/cutlass-sycl#356 for the GEMM accumulation in bf16. We can only merge this PR once codeplaysoftware/cutlass-sycl#356 has been merged and the |
Signed-off-by: Lukas Sommer <[email protected]>
FMarno
approved these changes
May 15, 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Outdated suggestions cannot be applied.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.
Enable autotuning for SYCL CUTLASS by completing the SYCL benchmark request class.
Also removes a temporary workaround that forced float32 accumulation to now allow GEMM to accumulate in bfloat16.
This addresses one of the items left open in #2.