Speed up permute propagation cleanup (#22161) - #22161
Open
apullin wants to merge 1 commit into
Open
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22161
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit d5cbd3e with merge base 4ce2ec2 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
Contributor
|
@apullin has exported this pull request. If you are a Meta employee, you can view the originating Diff in D114224762. |
This PR needs a
|
meta-codesync
Bot
changed the title
apullin
added a commit
to apullin/executorch
that referenced
this pull request
Aug 27, 2026
Summary: Speed up permute propagation cleanup. ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations: - Canonicalize view/permute chain collection now uses deque + membership set instead of list pop(0) and remove(), avoiding quadratic bookkeeping. - FuseDuplicateUsersPass deduplicates pending producer revisits while preserving same revisit behavior after fusions. - PropagateViewCopyPermutePass still retraces after each moved transform for metadata safety, but defers horizontal/vertical cleanup until full scan finds no more direct propagation moves. Preserves fixed-point behavior while avoiding expensive cleanup after every single moved transform. Behavior-preserving speedup, all existing pass tests pass. Differential Revision: D114224762
apullin
force-pushed
the
export-D114224762
branch
from
August 27, 2026 14:57
858c82b to
bb32005
Compare
apullin
added a commit
to apullin/executorch
that referenced
this pull request
Aug 27, 2026
Summary: Speed up permute propagation cleanup. ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations: - Canonicalize view/permute chain collection now uses deque + membership set instead of list pop(0) and remove(), avoiding quadratic bookkeeping. - FuseDuplicateUsersPass deduplicates pending producer revisits while preserving same revisit behavior after fusions. - PropagateViewCopyPermutePass still retraces after each moved transform for metadata safety, but defers horizontal/vertical cleanup until full scan finds no more direct propagation moves. Preserves fixed-point behavior while avoiding expensive cleanup after every single moved transform. Behavior-preserving speedup, all existing pass tests pass. Differential Revision: D114224762
apullin
force-pushed
the
export-D114224762
branch
from
August 27, 2026 15:21
bb32005 to
86beae5
Compare
apullin
added a commit
to apullin/executorch
that referenced
this pull request
Aug 28, 2026
Summary: Speed up permute propagation cleanup. ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations: - Canonicalize view/permute chain collection now uses deque + membership set instead of list pop(0) and remove(), avoiding quadratic bookkeeping. - FuseDuplicateUsersPass deduplicates pending producer revisits while preserving same revisit behavior after fusions. - PropagateViewCopyPermutePass still retraces after each moved transform for metadata safety, but defers horizontal/vertical cleanup until full scan finds no more direct propagation moves. Preserves fixed-point behavior while avoiding expensive cleanup after every single moved transform. Behavior-preserving speedup, all existing pass tests pass. On an ensemble network of ~1M parameters, lowering produced identical output while running 102.5 seconds faster (26.7%). Differential Revision: D114224762
apullin
force-pushed
the
export-D114224762
branch
from
August 28, 2026 15:57
86beae5 to
4985dfe
Compare
Contributor
|
This needs a rebase with some of the other refactors we've been doing to these passes. |
Contributor
Author
|
@rascani Sadly, it seems there is a bit of a paradoxical situation with the stuck diff train. I haven't been able to find a place where I can rebase to where export will work and conflicts will be solved externally and on the oss internal signal. But having the approvals in place is great bc I can queue this up if/when the internal diff train catches up. |
apullin
added a commit
to apullin/executorch
that referenced
this pull request
Sep 3, 2026
Summary: Speed up permute propagation cleanup. ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations: - Canonicalize view/permute chain collection now uses deque + membership set instead of list pop(0) and remove(), avoiding quadratic bookkeeping. - FuseDuplicateUsersPass deduplicates pending producer revisits while preserving same revisit behavior after fusions. - PropagateViewCopyPermutePass still retraces after each moved transform for metadata safety, but defers horizontal/vertical cleanup until full scan finds no more direct propagation moves. Preserves fixed-point behavior while avoiding expensive cleanup after every single moved transform. Behavior-preserving speedup, all existing pass tests pass. On an ensemble network of ~1M parameters, lowering produced identical output while running 102.5 seconds faster (26.7%). Differential Revision: D114224762
apullin
force-pushed
the
export-D114224762
branch
from
September 3, 2026 02:04
4985dfe to
d033cdf
Compare
Summary: Speed up permute propagation cleanup. ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations: - Canonicalize view/permute chain collection now uses deque + membership set instead of list pop(0) and remove(), avoiding quadratic bookkeeping. - FuseDuplicateUsersPass deduplicates pending producer revisits while preserving same revisit behavior after fusions. - PropagateViewCopyPermutePass still retraces after each moved transform for metadata safety, but defers horizontal/vertical cleanup until full scan finds no more direct propagation moves. Preserves fixed-point behavior while avoiding expensive cleanup after every single moved transform. Behavior-preserving speedup, all existing pass tests pass. On an ensemble network of ~1M parameters, lowering produced identical output while running 102.5 seconds faster (26.7%). Differential Revision: D114224762
apullin
force-pushed
the
export-D114224762
branch
from
September 6, 2026 15:06
d033cdf to
d5cbd3e
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:
Speed up permute propagation cleanup.
ARM lowering spends significant time in permute propagation. Reduce avoidable repeated work while keeping same optimizations:
Behavior-preserving speedup, all existing pass tests pass.
On an ensemble network of ~1M parameters, lowering produced identical output while running 102.5 seconds faster (26.7%).
Differential Revision: D114224762