Unify unrolled euclidean distance computation functions and unroll for int8_t
#32194
Conversation
This file contains 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
…r `int8_t`
Experiments on Mac M1 show the following unrolling results for `int8_t`:
1: 84 ms (baseline)
2: 41 ms
4: 53 ms
8: 82 ms
16: 120 ms
I.e. an unrolling factor of 2 is chosen for `int8_t`. It is expected that
this will also produce gains on x64 AVX{2,512}, but the concrete unrolling
factor has not been fine-tuned for this architecture yet.
baldersheim
approved these changes
Aug 20, 2024
vekterli
deleted the
vekterli/unify-unrolled-euclidean-distance-and-unroll-i8
branch
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Outdated suggestions cannot be applied.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.
@baldersheim please review 🔍👀
Experiments on Mac M1 show the following unrolling factor results for
int8_t:I.e. an unrolling factor of 2 is chosen for
int8_t. It is expected that this will also produce gains on x64 AVX{2,512}, but the concrete unrolling factor has not been fine-tuned for this architecture (yet).