Arm: Add SVE implementations of simdFilterHor #623
Merged
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Outdated suggestions cannot be applied.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.
Arm: Add SVE implementations of simdFilterHor
Add SVE implementations of the simdInterpolateHor filters for the 4-tap, 6-tap, and 8-tap.
Performance uplift vs Neon:
N8: 1.36x
N6: 1.09x
N4: 1.09x
These benchmarks are obtained from a Neoverse V2 using LLVM 21.
Arm: Refactor simdFilter_neon for width = 1 path
width == 1improves the average performance by 2.0x across all taps compared to the current scalar implementation.