More generic slice propagation before unary ops which works for non-contiguous slices (#19345)

[DrJessop]

Summary:

The existing MoveSliceToInputPass in the Jarvis compiler propagates slices backward through computation chains using the tiling/region infrastructure. However, it only supports contiguous slices (step=1) because non-unitary steps cannot be represented as contiguous regions.

This diff adds PropagateSlice, a lightweight pass that swaps slice_copy past element-wise quantize/dequantize ops when a cost model indicates it reduces data movement. Unlike MoveSliceToInputPass, it handles any step size (including non-contiguous slices like step=2), but only swaps past a single adjacent op rather than walking entire chains. The two passes are complementary: MoveSliceToInputPass handles deep propagation of contiguous slices through complex op chains with tiling, while PropagateSlice handles the simpler case of moving strided slices past quant/dequant boundaries where tiling is irrelevant (so far, but will be extended to other cases). The idea is eventually it can be applied iteratively to keep pushing the slice up.

Changes:

• Add PropagateSlice pass to reorder_ops.py with a dispatch-table design for extensibility
• Cost model: only swap when the slice actually reduces tensor volume (sliced < full)
• Supported ops: quantize_per_tensor, dequantize_per_tensor (both cadence and quantized_decomposed variants)
• Tests moved into test_reorder_ops_passes.py alongside other reorder pass tests

Reviewed By: ethansfng

Differential Revision: D103752840

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19345

• 📄 Preview Python docs built from this PR

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

❌ 30 Cancelled Jobs, 2 Unrelated Failures

As of commit b6f35de with merge base 851cffb ():

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• Check Labels (gh)
• periodic (gh)
• pull / android / run-emulator (gh)
• pull / test-coreml-bc-macos (macos-m1-stable) / macos-job (gh)
• pull / test-lora-linux / linux-job (gh)
• pull / test-models-linux (ic4, portable, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (ic4, xnnpack-quantization-delegation, linux.4xlarge.memory) / linux-job (gh)
• pull / test-models-linux (llama3_2_vision_encoder, portable, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (mobilebert, portable, linux.2xlarge) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (mobilebert, xnnpack-quantization-delegation, linux.2xlarge) / linux-job (gh)
• pull / test-models-linux (phi_4_mini, portable, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-moshi-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-phi-3-mini-runner-linux / linux-job (gh)
• pull / test-qnn-delegate-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-qnn-testsuite-linux / package-golden-artifacts (gh)
• pull / test-qnn-testsuite-linux / test-backend-linux (qnn, models) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-sqnr-static-llm-qnn-linux (smollm2_135m) / linux-job (gh)
• pull / test-voxtral-realtime-xnnpack-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-vulkan-models-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / unittest / linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / unittest / macos / macos-job (gh)
• pull / unittest-arm-backend-with-no-deps (test_pytest_models_tosa) / linux-job (gh)
• pull / unittest-arm-backend-with-no-deps (test_pytest_ops_tosa) / linux-job (gh)
• pull / unittest-buck / macos / macos-job (gh)
• pull / unittest-editable / linux / linux-job (gh)
• pull / unittest-editable / macos / macos-job (gh)
• pull / unittest-editable / windows / windows-job (gh)
• pull / unittest-nxp-neutron / linux-job (gh)
  ##[error]The operation was canceled.
• Test CoreML Backend / test-coreml / test-backend-macos (coreml, models) / macos-job (gh)
• Test CoreML Backend / test-coreml / test-backend-macos (coreml, operators) / macos-job (gh)

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

• pull / unittest / windows / windows-job (gh) (matched win rule in flaky-rules.json)
  ##[error]The operation was canceled.

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

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

• pull / test-lora-multimethod-linux / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.

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

[meta-codesync]

@DrJessop has exported this pull request. If you are a Meta employee, you can view the originating Diff in D103752840.

[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.