Remove no-op clones in xnnpack

[GregoryComer]

ATen clone ops can end up in the graph from a few sources. Since the graph is functional, we don't actually need these and they are slow. This PR runs the no-op clone removal pass for XNNPACK.

In addition to this, I ran into an issue where XNNPACK delegate doesn't currently handle inputs being directly forwarded to partition outputs. There has to be at least one operator. To solve this, I updated the removal pass to leave these clone ops in and added copy support in the XNN delegate to direct copy to the output.

In the long-run, I want to remove these no-ops higher up as part of to_edge, but this requires alignment and changes with a few more backends. See #15838. But resolving for XNNPACK will mitigate the issue for CPU models, at least.

Differential Revision: D87405074

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15884

• 📄 Preview Python docs built from this PR

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

✅ You can merge normally! (2 Unrelated Failures)

As of commit cafe8a3 with merge base aff5086 ():

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• pull / unittest / linux / linux-job (gh) (similar failure)
  exir/backend/test/test_lowered_backend_module.py::TestBackendAPI::test_emit_nested_lowered_backend_module

BROKEN TRUNK - The following job failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / unittest-editable / linux / linux-job (gh) (trunk failure)
  exir/backend/test/test_lowered_backend_module.py::TestBackendAPI::test_emit_nested_lowered_backend_module

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

[meta-codesync]

@GregoryComer has exported this pull request. If you are a Meta employee, you can view the originating Diff in D87405074.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[meta-codesync]

@GregoryComer has imported this pull request. If you are a Meta employee, you can view this in D87405074.

[Copilot]

Pull Request Overview

This PR adds support for removing no-op clone operations in the XNNPACK backend to improve performance, as clone operations that don't modify memory format or dim order add unnecessary overhead in functional graphs.

• Added RemoveCloneOpsTransform pass to the XNNPACK pass pipeline with support for preserving input-output copies
• Implemented clone/copy operator support in XNNPACK delegate for dim-order preserving clones
• Updated test models to include operations after clone to prevent direct input-to-output forwarding

Reviewed Changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
backends/transforms/remove_clone_ops.py	Added preserve_input_output_copies parameter and _is_input_output_copy method to selectively preserve clones that directly connect graph inputs to outputs
backends/xnnpack/_passes/init.py	Added RemoveCloneOpsTransform to the XNNPACK pass pipeline
backends/xnnpack/_passes/TARGETS	Added dependency on remove_clone_ops transform
backends/xnnpack/partition/config/generic_node_configs.py	Added CloneDimOrderConfig to partition dim-order preserving clones
backends/xnnpack/partition/config/init.py	Exported CloneDimOrderConfig
backends/xnnpack/operators/op_clone.py	Implemented CloneVisitor to serialize clone operations as XNN copy nodes
backends/xnnpack/operators/init.py	Added import for op_clone module
backends/xnnpack/serialization/xnnpack_graph_schema.py	Added XNNCopy node type to schema
backends/xnnpack/serialization/schema.fbs	Added XNNCopy to flatbuffer schema union
backends/xnnpack/serialization/runtime_schema.fbs	Added XNNCopy to runtime flatbuffer schema union
backends/xnnpack/runtime/XNNCompiler.cpp	Implemented defineCopyNode function to define XNN copy operations in the subgraph
backends/xnnpack/test/ops/test_clone.py	Added comprehensive tests for clone operation partitioning with various memory formats and tensor dimensions
exir/tests/test_memory_format_ops_pass_utils.py	Updated test models to include operations after clone to prevent direct forwarding

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.