Add initial permute-view fusing test-suite.#20415
Open
AdrianLundell wants to merge 2 commits into
Open
Conversation
Initial work on pytorch#20097, using the tests from the transpose count suite in the arm backend. All modules are exported and to_edge_transform_and_lowered with a single mock-pass which is intended to run the full pipeline for converting a graph to contiguous channels last format and fusing any additional permutes. The number of permutes and views are counted before and after the pipeline to track any change in the models. Other backends are encouraged to add their own test of interest to ensure that changes to the in-common passes will not regress individual backends. Additionally a number of full model tests will be added. Signed-off-by: Adrian Lundell <adrian.lundell@arm.com> Change-Id: I0c4cc5ba72036e539ab640ddaed2c63b145874cf
AdrianLundell
added
partner: arm
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20415
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: ❌ 2 New Failures, 3 Unrelated FailuresAs of commit 01c3c9b with merge base 0da9ca3 ( NEW FAILURES - The following jobs have failed:
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
There was a problem hiding this comment.
Pull request overview
Adds an initial (currently skipped) test suite under backends/transforms/test/ intended to track permute/view operator counts before/after running an edge transform/lower pipeline, and to verify numerical equivalence across a set of representative modules/models.
Changes:
- Introduces a catalog of test modules and “case” definitions covering conv/linear/attention/RNN/etc. plus channels-last variants.
- Adds a custom
ExportPassthat counts permute- and view-like Edge ops before/after the pass pipeline. - Adds two pytest entrypoints (normal + channels-last), currently marked as proof-of-concept and skipped.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| @pytest.mark.skip( | ||
| reason="Proof of concept - currently no permute-view passes implemented." | ||
| ) | ||
| @common.parametrize("case", cases_channels_last, xfail=xfails) |
| class MaxPool2dDilatedModule(torch.nn.Module): | ||
| def __init__(self): | ||
| super().__init__() | ||
| self.pool = torch.nn.MaxPool2d(kernel_size=3, stride=1, padding=1, dilation=2) |
Signed-off-by: Adrian Lundell <adrian.lundell@arm.com> Change-Id: Ibc5f48266d5015c0859be05d8eada92a84a682d5
zingo
approved these changes
Jun 22, 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
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.
Initial work on #20097, using the tests from the transpose count suite in the arm backend. All modules are exported and to_edge_transform_and_lowered with a single mock-pass which is intended to run the full pipeline for converting a graph to contiguous channels last format and fusing any additional permutes.
The number of permutes and views are counted before and after the pipeline to track any change in the models.
Other backends are encouraged to add their own test of interest to ensure that changes to the in-common passes will not regress individual backends. Additionally a number of full model tests will be added.
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani