Don't cache source partitions

[lucylq]

Summary

This caching breaks multi-method; for the second method, we are using the graph of the first method, and do not delegate properly. I think this is because the source partitions are based on the first graph, so the addmm nodes do not match up/do not give a partition, so the graph doesn't get tagged by the partitioner.

See issue: #12804

In the example from the issue, the first method fwd is delegated well, and the second method fwd_2 does not have the linear functions delegated (it is using portable kernels, and the weights are duplicated in the constant segment).

Also add test_partitioner to buck targets.

Test plan

python -m unittest backends.xnnpack.test.test_xnnpack_partitioner

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15541

• 📄 Preview Python docs built from this PR

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

❌ 8 New Failures, 4 Unrelated Failures

As of commit 32b4794 with merge base cadb0db ():

NEW FAILURES - The following jobs have failed:

• Test CUDA Builds / test-model-cuda-e2e (google, gemma-3-4b-it, non-quantized) / linux-job (gh)
  RuntimeError: Command docker exec -t 6381c6c08771dbb131987d1b40e17b1bffaba6581960c1f7ef487b61e5fea8c7 /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (google, gemma-3-4b-it, quantized-int4-tile-packed) / linux-job (gh)
  RuntimeError: Command docker exec -t d0c5dcc261296b89c17d1ec1b600bb075d3a23a6bdd1cec69b2f62fcaffb3b11 /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-large-v3-turbo, non-quantized) / linux-job (gh)
  RuntimeError: Command docker exec -t f6c3acc8027fdaca7bd7f5cdb79ab0e9586db96fec5bf887cd355c5d64f617ed /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-large-v3-turbo, quantized-int4-tile-packed) / linux-job (gh)
  RuntimeError: Command docker exec -t 46470a2edd7a6999a56af43aab0f1c1db892b2c17ee81fa9f981a9e456b3289a /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-large-v3-turbo, quantized-int4-weight-only) / linux-job (gh)
  RuntimeError: Command docker exec -t f609a231d64413609c8e799e20f0849dba67c8b0eaff827004c8d5905e4f96ad /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-small, non-quantized) / linux-job (gh)
  RuntimeError: Command docker exec -t 8cd793c325f63654223699d921432a2d2328ae58d3fdc6880cd534527bf10fa2 /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-small, quantized-int4-tile-packed) / linux-job (gh)
  RuntimeError: Command docker exec -t af70791ef340e77446638ad641fbcb9c5e6dcdad62ae08da714b8e596d5795cc /exec failed with exit code 1
• Test CUDA Builds / test-model-cuda-e2e (openai, whisper-small, quantized-int4-weight-only) / linux-job (gh)
  RuntimeError: Command docker exec -t 8446335a47cdd60bf3a4f5303d5c6e952d4ff26426f47e1c7eada9dba370b26a /exec failed with exit code 1

BROKEN TRUNK - The following jobs failed but were present on the merge base:

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

• pull / unittest / linux / linux-job (gh) (trunk failure)
  backends/xnnpack/test/recipes/test_xnnpack_recipes.py::TestXnnpackRecipes::test_int8_static_quant_recipe
• pull / unittest / macos / macos-job (gh) (trunk failure)
  backends/xnnpack/test/recipes/test_xnnpack_recipes.py::TestXnnpackRecipes::test_int8_static_quant_recipe
• pull / unittest-editable / linux / linux-job (gh) (trunk failure)
  backends/xnnpack/test/recipes/test_xnnpack_recipes.py::TestXnnpackRecipes::test_int8_static_quant_recipe
• pull / unittest-editable / macos / macos-job (gh) (trunk failure)
  backends/xnnpack/test/recipes/test_xnnpack_recipes.py::TestXnnpackRecipes::test_int8_static_quant_recipe

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

[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]

@lucylq has imported this pull request. If you are a Meta employee, you can view this in D86149805.

[GregoryComer]

Thanks for adding this. FYI We've deprecated the to_edge + to_backend flow for XNNPACK and will throw up a big warning when used. Fixing the caching issue is probably good regardless, even if it mostly only affects to_edge + to_backend (we preserve linear w/ to_edge_transform_and_lower).