Add DecomposeChannelsLastPass for the channels_last CPU fallback#20615
Merged
Conversation
Stacked on the channels_last dialect (pytorch#20093). Provides the channels-first CPU fallback for the dialect: an ExportPass that rewrites channels_last.{convolution,avg_pool2d,permute_copy} into permute_copy + edge.aten op + permute_copy, running on the existing portable kernels with memory-planned intermediates. Intended for the graph portion that no backend claimed; backends instead replace the channels_last ops with their own channels-last kernels. This is the decomposition alternative to a C++ reference kernel: per ExecuTorch precedent, decomposing the un-delegated remainder via a post-partition pass is idiomatic, whereas a kernel dispatching to the portable ops (registry or forward-declared symbols) has no in-tree precedent and needs unplanned scratch. Part of pytorch#20093. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20615
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit a3bb536 with merge base f4a6aac ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
AdrianLundell
approved these changes
Jun 30, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Provides the channels-first CPU fallback for the dialect: an ExportPass that rewrites channels_last.{convolution,avg_pool2d,permute_copy} into permute_copy + edge.aten op + permute_copy, running on the existing portable kernels with memory-planned intermediates. Intended for the graph portion that no backend claimed; backends instead replace the channels_last ops with their own channels-last kernels.
Part of #20093.
Test plan