Arm backend: Reuse identical CONST_SHAPE nodes#19770
Merged
Merged
Conversation
Cache CONST_SHAPE nodes created by InsertConstShapesPass and reuse them when a later view/repeat needs the same shape. This removes duplicate shape constants. This improvement is model dependent. Models with few repeated literal shapes will not see any meaningful change, but some models can benefit from it notably. The table below shows the results of a local test lowering DeiT Tiny to TOSA-FP. The lowering time reduced in this run, likely because passes following InsertConstShapesPass had fewer nodes to iterate over. | Metric | Baseline | Optimized | Delta | | -------------- | -------- | --------- | ---------------- | | Total ops | 2106 | 1736 | -370 (-17.6%) | | CONST_SHAPE | 466 | 96 | -370 (-79.4%) | | TOSA size | 23.82 MB | 23.75 MB | -71.6 KB (-0.3%) | | Execution time | 118.7 s | 78.4 s | -40.3 s (-34.0%) | Signed-off-by: Martin Lindström <Martin.Lindstroem@arm.com> Change-Id: I620b04e496996fafebc9c7f7854fd8399b3d6d4b
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19770
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ⏳ 38 Pending, 1 Unrelated FailureAs of commit d06f7fd with merge base 77df9b7 ( BROKEN TRUNK - The following job 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
Collaborator
Author
|
@pytorchbot label ciflow/trunk |
Collaborator
Author
|
@pytorchbot label "partner: arm" |
pytorch-bot
Bot
added
the
partner: arm
Collaborator
Author
|
@pytorchbot label "release notes: none" |
pytorch-bot
Bot
added
the
release notes: none
zingo
approved these changes
May 26, 2026
zingo
approved these changes
May 26, 2026
Contributor
There was a problem hiding this comment.
Pull request overview
This PR optimizes the Arm backend lowering pipeline by reusing identical tosa.CONST_SHAPE nodes emitted by InsertConstShapesPass, reducing duplicate shape constants in generated TOSA graphs.
Changes:
- Adds an instance-level cache keyed by shape tuple.
- Reuses previously created
CONST_SHAPEnodes for laterview_copy/repeatops with the same literal shape.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+54
to
+55
| const_node = self._const_shape_cache.get(shape) | ||
| if const_node is None: |
Comment on lines
+54
to
+55
| const_node = self._const_shape_cache.get(shape) | ||
| if const_node is None: |
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
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.
Cache CONST_SHAPE nodes created by InsertConstShapesPass and reuse them when a later view/repeat needs the same shape. This removes duplicate shape constants.
This improvement is model dependent. Models with few repeated literal shapes will not see any meaningful change, but some models can benefit from it notably.
The table below shows the results of a local test lowering DeiT Tiny to TOSA-FP. The lowering time reduced in this run, likely because passes following InsertConstShapesPass had fewer nodes to iterate over.
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani