Add NEON fast path for SFP weight decode - #971
Open
RaviTriv wants to merge 3 commits into
Open
Conversation
RaviTriv
marked this pull request as ready for review
August 4, 2026 00:25
jan-wassenberg
approved these changes
Aug 5, 2026
jan-wassenberg
left a comment
jan-wassenberg
left a comment
Member
There was a problem hiding this comment.
Nice work and notable speedup, thanks for adding :) Have you measured on non-AVX3_DL x86 to see if your approach is faster than the generic? If so, we could replace our generic fallback on all platforms, rather than just NEON. SVE is also increasingly available/chosen.
jan-wassenberg
added
the
copybara-import
Author
thanks! good idea, have not measured. taking a look :) |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Proposed fast path for weight decode on NEON that uses table look up and shift. About the same ops but looks like we can sneak in a few cheaper ones, roughly 1.1x faster on SFP matmuls.