[ExecuTorch][WebGPU] mul op test suite (cases.py op-test framework)

[JulianCloudNTH]

Stack from ghstack (oldest at bottom):

• [ExecuTorch][WebGPU] aten.index.Tensor test suite (export + native golden) #20465
• [ExecuTorch][WebGPU] Add aten.index.Tensor (1D-self gather) #20464
• [ExecuTorch][WebGPU] Add clone op (aten.clone.default) #20463
• [ExecuTorch][WebGPU] Flatten landed-op test dirs to test/ops/test_<op>.py #20435
• [ExecuTorch][WebGPU] cat op test suite (cases.py op-test framework) #20399
• [ExecuTorch][WebGPU] Add cat op + ValueList graph support (aten.cat.default) #20398
• [ExecuTorch][WebGPU] permute_copy op test suite (cases.py op-test framework) #20397
• [ExecuTorch][WebGPU] Add permute_copy + IntList graph support (aten.permute_copy.default) #20396
• [ExecuTorch][WebGPU] slice_copy op test suite (cases.py op-test framework) #20395
• [ExecuTorch][WebGPU] Add slice_copy op (aten.slice_copy.Tensor) #20394
• [ExecuTorch][WebGPU] squeeze_copy + unsqueeze_copy test suites (cases.py op-test framework) #20393
• [ExecuTorch][WebGPU] Add squeeze_copy + unsqueeze_copy (flat copies) #20392
• [ExecuTorch][WebGPU] sigmoid op test suite (cases.py op-test framework) #20391
• [ExecuTorch][WebGPU] Add sigmoid op (aten.sigmoid.default) #20390
• [ExecuTorch][WebGPU] select_copy op test suite (cases.py op-test framework) #20363
• [ExecuTorch][WebGPU] Add select_copy op (aten.select_copy.int) #20362
• [ExecuTorch][WebGPU] view_copy op test suite (cases.py op-test framework) #20361
• [ExecuTorch][WebGPU] Add view_copy op (aten.view_copy.default) #20360
• -> [ExecuTorch][WebGPU] mul op test suite (cases.py op-test framework) #20359

Registers aten.mul.Tensor in the cases.py op-test framework: a _mul_suite of 3 configs (same-shape fast path, last-dim broadcast at LLM width, mixed-rank left-pad) that generate_op_tests exports via VulkanPartitioner and compares to an fp64 torch golden on Dawn. Also adds test/ops/mul/test_mul.py (MulModule + CONFIGS + an export-delegation/eager smoke test) and the aten.mul.Tensor partitioner-allowlist entry in tester.py.
@exported-using-ghexport

Differential Revision: D108793154

Differential Revision: D108793154

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20359

• 📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 2 Unrelated Failures

As of commit f8f40bb with merge base e03f777 ():

NEW FAILURE - The following job has failed:

• pull / unittest-buck / linux / linux-job (gh)
  RuntimeError: Command docker exec -t b8a088f086ad9d0acac839b7ad0ecd2869bf03f0bd6d737fab0a2fc8cad4de09 /exec failed with exit code 3

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

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

• pull / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh) (trunk failure)
• pull / unittest-buck / macos / macos-job (gh) (trunk failure)
  RuntimeError: Command bash /Users/ec2-user/runner/_work/_temp/exec_script failed with exit code 3

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.

[JulianCloudNTH]

@claude review

[claude]

Claude finished @JulianCloudNTH's task in 1m 46s —— View job

---

Code Review — WebGPU mul op test suite

• Read PR diff
• Review against add / rms_norm conventions
• Check cases.py framework integration & runtime (BinaryOp.cpp)
• Post review

---

Overall this is a clean, well-scoped addition that fits the existing op-test framework. The three configs are well-chosen (same-shape fast path, last-dim broadcast at LLM width, mixed-rank left-pad) and exercise the broadcast paths in runtime/ops/mul/BinaryOp.cpp (fill_tensor_meta_broadcast). The tester.py allowlist entry and cases.py registration are correct. A few notes below, mostly minor.

1. test_golden_matches_eager is tautological (low priority)

backends/webgpu/test/ops/mul/test_mul.py:66-69

def test_golden_matches_eager(self) -> None:
    for _, (sa, sb) in CONFIGS.items():
        a, b = _det_inputs(sa, sb)
        torch.testing.assert_close(MulModule()(a, b), a * b)

MulModule.forward is a * b, so this asserts a * b == a * b — it always passes and verifies no real reference. It does smoke-test that the module instantiates and broadcasts without error, but the docstring's "eager-correctness" framing oversells it. The actual numeric oracle is the fp64 golden compared on Dawn by the cases.py suite, so this assertion is redundant. Consider either dropping the numeric assertion (keep it as a "module runs + shapes broadcast" check) or renaming to reflect that it's a shape/smoke test. Fix this →

2. Import path differs from sibling tests (style nit)

test_mul.py:18 uses:

from executorch.backends.vulkan import VulkanPartitioner

while test_add.py and test_rms_norm.py use the fully-qualified path:

from executorch.backends.vulkan.partitioner.vulkan_partitioner import VulkanPartitioner

Both resolve (the package __init__ re-exports it), but matching the siblings keeps the directory consistent.

3. module_factory=lambda: MulModule() could just be MulModule

cases.py:119 — since no Case here sets construct, the factory is always called with no kwargs, so passing the class directly (module_factory=MulModule) is equivalent and slightly simpler. The lambda is harmless; only worth changing if you prefer the brevity.

Minor observations (no action needed)

• _det_inputs reuses one Generator(seed=0) sequentially for a then b, so they get distinct values — correct. Note these deterministic inputs are only used by the local smoke tests; the cases.py suite feeds bare shape tuples, which default to randn per InputSpec.
• The mixedrank case ((4,), (3, 4)) correctly covers in.ndim < out.ndim, and stays within the rank-4 guard in BinaryOp.cpp:37-41. Good coverage choice.
• Copyright headers, docstrings, and __init__.py all match repo conventions.

Nothing blocking here — the main thing worth addressing is the vacuous assertion in (1).
· branch gh/JulianCloudNTH/34/head