perf(avx512f): single vpermt2 for zip_lo/zip_hi#1370
Merged
DiamonDinoia merged 1 commit intoJul 6, 2026
Conversation
The AVX-512 zip_lo/zip_hi lowered the interleave as per-128-bit unpacklo/unpackhi followed by a 3x insertf32x4 / extractf32x4 lane-fixup pile (~7 shuffle-port uops) needed to undo AVX-512's in-lane element ordering. A single 2-source cross-lane permute (vpermt2d/vpermt2q for integral, vpermt2ps/vpermt2pd for float/double) produces the same result in one port-5 uop.
Contributor
|
This is much better. Thanks for looking at this. |
serge-sans-paille
approved these changes
Jul 6, 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The AVX-512 zip_lo/zip_hi lowered the interleave as per-128-bit unpacklo/unpackhi followed by a 3x insertf32x4 / extractf32x4 lane-fixup pile (~7 shuffle-port uops) needed to undo AVX-512's in-lane element ordering. A single 2-source cross-lane permute (vpermt2d/vpermt2q for integral, vpermt2ps/vpermt2pd for float/double) produces the same result in one port-5 uop.