[CUBLAS] Enable offloading of R.matmul + R.dequantize#16896
Merged
masahi merged 1 commit intoapache:mainfrom May 3, 2024
Merged
[CUBLAS] Enable offloading of R.matmul + R.dequantize#16896masahi merged 1 commit intoapache:mainfrom
masahi merged 1 commit intoapache:mainfrom
Conversation
This commit enables offloading of R.matmul + R.dequantize to cuBLAS codegen. Dequantization scale is passed to runtime function and set to alpha parameter. If there is no dequantization, then alpha == 1.0.
ibsidorenko
force-pushed
the
cublas-matmul-dequantize
branch
from
df3b24f to
40d6d90
Compare
masahi
approved these changes
May 3, 2024
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This commit enables offloading of
R.matmul+R.dequantizeto cuBLAS codegen. Dequantization scale is passed to runtime function and set to alpha parameter. If there is no dequantization, then alpha == 1.0.Also, it can be used to fuse output scale to matmul in case of FP8.
cc @vinx13 @csullivan @JosephTheOctonaut @masahi @elvin-n