Implement reduce_product() - #361
Merged
Merged
Conversation
…o 33% boost for 16-bit
LaurenzV
approved these changes
Sep 3, 2026
Comment on lines
+3632
to
+3638
| let mul_level_0: [i32; 2usize] = [ | ||
| a[0usize].wrapping_mul(a[1usize]), | ||
| a[2usize].wrapping_mul(a[3usize]), | ||
| ]; | ||
| let mul_level_1: [i32; 1usize] = [mul_level_0[0usize].wrapping_mul(mul_level_0[1usize])]; | ||
| mul_level_1[0] | ||
| } |
Collaborator
There was a problem hiding this comment.
I'm wondering whether the compiler will actually compile this just into a sequence of multiplications or whether there will be loads/stores?
Contributor
Author
There was a problem hiding this comment.
Since all these are local stack values with very clearly no aliasing, the intermediate stores should in theory be eliminated.
The reason it's emitted like this with consecutive arrays is to make autovectorization easier on platforms we don't already support via intrinsics.
# Conflicts: # fearless_simd_gen/src/mk_wasm.rs # fearless_simd_gen/src/mk_x86.rs
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Matches
std::simdin precision (same number of roundings), but uses a more SIMD-friendly multiplication order. It uses the same order across all backends. The implementation is similar toreduce_sum().The result is not guaranteed to be equal to
v.to_slice().sum(), but that is only achievable with fully scalar multiplication;std::simdscalarizes to uphold this behavior which it doesn't even document.Unlike #357 we can't reduce precision loss to log2(N) rather than N, at least not without complex algorithms that compute the precise sum in one rounding by separately accounting for the accumulated error. A precise variant which does that could be added later.
This is the final part of #340.