Add Neon implementation of search_n#6108
Merged
StephanTLavavej merged 3 commits intomicrosoft:mainfrom Feb 28, 2026
Merged
Conversation
AlexGuteniev
approved these changes
Feb 25, 2026
StephanTLavavej
approved these changes
Feb 25, 2026
Member
|
I'm mirroring this to the MSVC-internal repo. Please notify me if any further changes are pushed, otherwise no action is required. |
Member
🦾 🕵️ 🖖 |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 PR adds a Neon implementation of
search_n, for small values ofn.The notable difference in approach to the SSE4.2/AVX2 paths is that here we use a 64-bit nibble mask, rather than a packed bitmask, in order to save on expensive MOVEMASK emulation. As a result of this, we handle the Carry via CLZ rather than mask concatenation.
I also included a 64-bit path, as this did actually seem to be a (modest) gain.
Results are a little mixed and variable, and contain some 'spurious' results for
n > 8where the vector implementation is not hit but the differences in the header implementation and the fallback are being measured instead. However they seem to roughly match he results for SSE4.2 in #5544.Benchmark results 🕐: