[EXIR] Register _clone_dim_order op and map aten.clone #13735

keyprocedure · 2025-08-27T20:27:05Z

Summary

This is PR 2 of 3 implementing a dim order aware clone op.

This PR registers the new _clone_dim_order op and maps aten.clone to dim_order_ops._clone_dim_order in EXIR during export to preserve memory layout changes (contiguous/channels_last). It also updates the Core ML, ARM, and Qualcomm backends to handle the new clone op.

Related PRs:

PR 1: #12974 - Add _clone_dim_order portable kernel
PR 3: #12976 - Update RemoveCloneOpsTransform to be dim order aware

Fixes #12645

Test plan

Operator level tests to verify kernel behavior for layout preservation and changes.
Graph level checks to confirm that clone mapping occurs.
End to end tests to validate that functional clone behavior is unchanged.
Backend tests to ensure clone semantics are preserved.

All tests pass via:
python -m unittest exir.tests.test_memory_format_ops_pass
python -m unittest backends.apple.coreml.test.test_torch_ops
pytest backends/arm/test/ops/test_clone.py
pytest backends/arm/test/passes/test_remove_clone_pass.py

pytorch-bot · 2025-08-27T20:27:08Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13735

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 3 New Failures

As of commit 30a1d13 with merge base 32e82bc ():

NEW FAILURES - The following jobs have failed:

Build documentation / build (buck2) / Build doc (gh)
At least one of the pre-conditions you specified did not hold
pull / unittest / macos / macos-job (gh)
backends/xnnpack/test/ops/test_rsqrt.py::TestRsqrt::test_fp16_rsqrt
trunk / test-selective-build-macos (cmake) / macos-job (gh)
RuntimeError: Command bash /Users/ec2-user/runner/_work/_temp/exec_script failed with exit code 1

This comment was automatically generated by Dr. CI and updates every 15 minutes.

keyprocedure · 2025-08-27T20:27:56Z

@pytorchbot label "release notes: none"

keyprocedure · 2025-09-03T19:25:42Z

@Gasoonjia I could use your input on the Qualcomm backend changes:

After replacing edge.aten.clone with edge.dim_order_ops._clone_dim_order in QNN, I was able to successfully run all the previously failing QNN tests on a local Linux machine.

Should _clone_dim_order always be treated as a non-delegated op, like clone was?

I set the clone op removal condition to the default value (True) instead of checking if dim order is non-contiguous.

I also didn't create a node visitor or add an operator test in test_qnn_delegate since the op wouldn't be delegated and would fail the delegated partitioner count test.

Gasoonjia · 2025-09-03T20:03:35Z

Hi @keyprocedure , thanks for your great work and ci has been triggered!

Yes we can treat the dim order variant copy as non-delegated, just as what we are doing for aten::copy.

cc. @digantdesai

keyprocedure · 2025-09-03T20:08:18Z

Happy to help!
Sounds good, really appreciate your responsiveness.

keyprocedure · 2025-09-03T23:30:27Z

@Gasoonjia CI failures are unrelated. If everything looks good to you, I can commit the test_torch_ops.py merge conflict fix.

Gasoonjia · 2025-09-04T04:17:06Z

@keyprocedure lgtm. Please solve the conflict and then I will stamp it!

keyprocedure · 2025-09-04T06:10:15Z

@Gasoonjia the merge conflict is resolved.

I noticed that coreml's torch_ops.py has updated the way it registers _empty_dim_order and _to_dim_order_copy. Since _clone_dim_order hasn't been registered in Apple's dim_order_ops.py yet, are the current changes in torch_ops.pysufficient for now?

pytorch-bot · 2025-09-04T06:51:46Z

To add the ciflow label ciflow/trunk please first approve the workflows that are awaiting approval (scroll to the bottom of this page).

This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows.

pytorch-bot · 2025-09-04T06:52:06Z

To add the ciflow label ciflow/trunk please first approve the workflows that are awaiting approval (scroll to the bottom of this page).

This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows.

Gasoonjia · 2025-09-04T06:54:55Z

@keyprocedure i think it is fine for us right now, and we can work on coreml side later.

@metascroy do you have any concerns on this?

metascroy · 2025-09-04T17:29:26Z

backends/apple/coreml/compiler/torch_ops.py

        to(context, node)


+@register_torch_op(


metascroy · 2025-09-04T17:32:28Z

@Gasoonjia the merge conflict is resolved.

I noticed that coreml's torch_ops.py has updated the way it registers _empty_dim_order and _to_dim_order_copy. Since _clone_dim_order hasn't been registered in Apple's dim_order_ops.py yet, are the current changes in torch_ops.pysufficient for now?

Yes, the changes in torch_ops.py should be sufficient. The purpose of torch_ops.py is to register ops that are not yet in coremltools. With that said, we generally prefer that you put up a PR in coremltools and link that PR in a comment above the op you're creating in toch_ops.py. You can see this for most other ops in the file, e.g.,

executorch/backends/apple/coreml/compiler/torch_ops.py

Line 39 in 82991b5

# https://github.com/apple/coremltools/pull/2557

Gasoonjia

Thanks @metascroy 's comment! Let me stamp this PR for now and we can add coreml PR later.

### Summary This is PR 3 of 3 implementing a dim order aware clone op. This PR updates the clone removal pass to retain layout changing `aten.clone` and `_clone_dim_order` ops and remove no-op clones, ensuring layout/dim order is preserved through export. Related PRs: - PR 1: [#12974](#12974) - Add `_clone_dim_order` portable kernel - PR 2: [#13735](#13735) - Register `_clone_dim_order` op and map `aten.clone` Fixes #12645 ### Test plan Added tests to verify: - Clones that change layout are preserved. - Clones with unchanged layout (identity ops) are removed. All tests pass via: python -m unittest exir.tests.test_memory_format_ops_pass python -m unittest backends.transforms.test.test_remove_clone_ops --------- Co-authored-by: Gasoonjia <[email protected]>

keyprocedure and others added 21 commits July 29, 2025 10:49

Register clone_dim_order op; add test for op replacement

877119f

Rename clone_dim_order op registration test

f75845d

Merge branch 'main' into add-dim-order-clone-aot

83d8c75

Add graph level and end to end tests for _clone_dim_order op

cff39c9

Remove _clone_dim_order op registration (moved to PR pytorch#12974)

1fe461f

Register _clone_dim_order op

95db027

Merge branch 'main' into add-dim-order-clone-aot

63d45e7

Merge branch 'main' into add-dim-order-clone-aot

d9a181c

Register _clone_dim_order as no-op in CoreML

c7caa27

Remove redundant _clone_dim_order graph check

e54605f

Add _clone_dim_order to RemoveClonePass and update op name in tests

246bc44

Register _clone_dim_order under TOSA support check

c48467c

Merge branch 'main' into add-dim-order-clone-aot

e262a36

Add clone_dim_order_support to TOSA operator support list

5546360

Register node visitor for _clone_dim_order

5c5e65a

Merge branch 'main' into add-dim-order-clone-aot

fe7dd11

Remove visitor node registration for _clone_dim_order

7a0bc6a

Remove aten.clone check from RemoveClonePass

74e2cce

Remove input dtype gating and add memory_format check

f9f9515

Merge branch 'main' into add-dim-order-clone-aot

8d0cb06

Add Core ML test for _clone_dim_order

6839212

keyprocedure requested review from JacobSzwejbka, cccclai, digantdesai, larryliu0820 and shoumikhin as code owners August 27, 2025 20:27

meta-cla bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Aug 27, 2025

pytorch-bot bot added the release notes: none Do not include this in the release notes label Aug 27, 2025

pytorch-bot bot removed the ciflow/trunk label Sep 3, 2025

Gasoonjia added the ciflow/trunk label Sep 3, 2025

Replace edge.aten.clone with edge.dim_order_ops._clone_dim_order

5bfd58b

pytorch-bot bot removed the ciflow/trunk label Sep 3, 2025

Gasoonjia added the ciflow/trunk label Sep 3, 2025

Merge branch 'main' into add-dim-order-clone-aot

f8ab347

pytorch-bot bot removed the ciflow/trunk label Sep 4, 2025

Fix formatting

4a2383b

Gasoonjia added the ciflow/trunk label Sep 4, 2025

pytorch-bot bot removed the ciflow/trunk label Sep 4, 2025

Gasoonjia added the ciflow/trunk label Sep 4, 2025

pytorch-bot bot removed the ciflow/trunk label Sep 4, 2025

Merge branch 'main' into add-dim-order-clone-aot

30a1d13

Gasoonjia added the ciflow/trunk label Sep 4, 2025

metascroy reviewed Sep 4, 2025

View reviewed changes

Gasoonjia approved these changes Sep 4, 2025

View reviewed changes

Gasoonjia merged commit c5ff74c into pytorch:main Sep 4, 2025
299 of 303 checks passed

This was referenced Sep 8, 2025

Add _clone_dim_order portable kernel #12974

Merged

[EXIR] Update RemoveCloneOpsTransform to be dim order aware #12976

Merged

[EXIR] Register _clone_dim_order op and map aten.clone #13735

[EXIR] Register _clone_dim_order op and map aten.clone #13735

Uh oh!

Conversation

keyprocedure commented Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Uh oh!

pytorch-bot bot commented Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13735

❌ 3 New Failures

Uh oh!

keyprocedure commented Aug 27, 2025

Uh oh!

keyprocedure commented Sep 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Gasoonjia commented Sep 3, 2025

Uh oh!

keyprocedure commented Sep 3, 2025

Uh oh!

keyprocedure commented Sep 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Gasoonjia commented Sep 4, 2025

Uh oh!

keyprocedure commented Sep 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Sep 4, 2025

Uh oh!

pytorch-bot bot commented Sep 4, 2025

Uh oh!

Gasoonjia commented Sep 4, 2025

Uh oh!

metascroy Sep 4, 2025

Choose a reason for hiding this comment

Uh oh!

metascroy commented Sep 4, 2025

Uh oh!

Gasoonjia left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

keyprocedure commented Aug 27, 2025 •

edited

Loading

pytorch-bot bot commented Aug 27, 2025 •

edited

Loading

keyprocedure commented Sep 3, 2025 •

edited

Loading

keyprocedure commented Sep 3, 2025 •

edited

Loading

keyprocedure commented Sep 4, 2025 •

edited

Loading