force neon vectorization for ios (#21479) - #21479
Open
JakeStevens wants to merge 1 commit into
Open
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21479
Note: Links to docs will display an error until the docs builds have been completed. ❌ 32 New Failures, 1 Cancelled Job, 10 Unrelated FailuresAs of commit 63d8cc0 with merge base 6bbb75a ( NEW FAILURES - The following jobs have failed:
CANCELLED JOB - The following job was cancelled. Please retry:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
Contributor
|
@JakeStevens has exported this pull request. If you are a Meta employee, you can view the originating Diff in D113940046. |
This PR needs a
|
meta-codesync
Bot
changed the title
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 30, 2026
Summary: The default implementation auto-vectorized well for Android builds, but the specific LLVM Clang 21 toolchain + compiler flags did not auto-vectorize, resulting in only scalar operations. This commit selectively forces neon vectorization for iOS Differential Revision: D113940046
JakeStevens
force-pushed
the
export-D113940046
branch
from
July 30, 2026 00:03
530ec25 to
57e01dc
Compare
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 30, 2026
Summary: The default implementation auto-vectorized well for Android builds, but the specific LLVM Clang 21 toolchain + compiler flags did not auto-vectorize, resulting in only scalar operations. This commit selectively forces neon vectorization for iOS Reviewed By: YIWENX14 Differential Revision: D113940046
JakeStevens
force-pushed
the
export-D113940046
branch
from
July 30, 2026 18:50
57e01dc to
e3b5b27
Compare
|
|
||
| ET_INLINE void vec_scale(float* data, float scale, int64_t size) { | ||
| int64_t idx = 0; | ||
| #if defined(__APPLE__) && defined(__ARM_NEON) |
Contributor
There was a problem hiding this comment.
I dont super love the #ifdefs throughout the kernel code I would rather have a singular layer in the kernel process that does dispatching similar to the torchao arm neon kernels we have. We can take that as a BE followup though
Contributor
Author
There was a problem hiding this comment.
have an internal task tracking, will fix when I also rename some stuff in a diff that supports scalar gated too
JacobSzwejbka
approved these changes
Aug 4, 2026
| state_row[v_idx] = updated; | ||
| output_t[v_idx] += updated * query_value; | ||
| int64_t idx = 0; | ||
| #if defined(__APPLE__) && defined(__ARM_NEON) |
Contributor
There was a problem hiding this comment.
why only apple? Is the auto vectorization on android just better then the handwritten intrinsics here?
Contributor
Author
There was a problem hiding this comment.
yes, at least with our internal setup. This is why we were seeing such a gap initially, Android autovectorization worked really well, Apple did not at all. This attempt at forcing it improved for Apple (going scalar -> vectored) but actually regressed for Android.
Summary: The default implementation auto-vectorized well for Android builds, but the specific LLVM Clang 21 toolchain + compiler flags did not auto-vectorize, resulting in only scalar operations. This commit selectively forces neon vectorization for iOS Reviewed By: YIWENX14, JacobSzwejbka Differential Revision: D113940046
JakeStevens
force-pushed
the
export-D113940046
branch
from
August 4, 2026 16:24
e3b5b27 to
63d8cc0
Compare
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.
Summary:
The default implementation auto-vectorized well for Android builds, but the specific LLVM Clang 21 toolchain + compiler flags did not auto-vectorize, resulting in only scalar operations.
This commit selectively forces neon vectorization for iOS
Reviewed By: YIWENX14, JacobSzwejbka
Differential Revision: D113940046