[ET-VK][ez] Align SDPA attention weights S dim to the next multiple of 4

[SS-JIA]

Stack from ghstack (oldest at bottom):

• [ET-VK][ez] Don't copy zeros for cache tensors #15580
• [ET-VK][ez] SDPA don't branch based on whether bounds check needed #15579
• -> [ET-VK][ez] Align SDPA attention weights S dim to the next multiple of 4 #15578
• [ET-VK][ez] Update SDPA test to be able to test different SDPA modes #15577
• [ET-VK][ez] Refactor yaml configs for SDPA shaders #15576

Title says it all!

Why?

• Technically, this is should not be needed but SDPA op was producing incorrect output on Samsung S24 with buffer input tensors. The exact root cause is unclear, but it appears to be an issue specific to the Adreno 750 since it does not reproduce on any other GPU. The best guess at the moment is that we need to ensure that there is no possibility of multiple threads writing to the same memory location.

Differential Revision: D86226134

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15578

• 📄 Preview Python docs built from this PR

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

❌ 4 New Failures, 5 Unrelated Failures

As of commit 7b097a3 with merge base bde6b11 ():

NEW FAILURES - The following jobs have failed:

• pull / unittest-arm-backend-with-no-fvp (test_pytest_ops) / linux-job (gh)
  RuntimeError: Command docker exec -t f0fc53f237506bff92ee3556631d010eb1422d196128f2de55ce6b99cbcde0c6 /exec failed with exit code 1
• Test CUDA Builds / export-model-cuda-artifact (google, gemma-3-4b-it, non-quantized) / linux-job (gh)
  RuntimeError: Command docker exec -t 66a66044234391b06f69c601417feeb18a4d56d559a81e20aec0f33fb2b41ffe /exec failed with exit code 1
• Test CUDA Builds / export-model-cuda-artifact (openai, whisper-small, non-quantized) / linux-job (gh)
  RuntimeError: No CUDA GPUs are available
• Test CUDA Builds / test-models-cuda (linear) / linux-job (gh)
  RuntimeError: No CUDA GPUs are available

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-binary-size-linux-gcc / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / test-openvino-linux / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / test-setup-linux-gcc / linux-job (gh) (trunk failure)
• pull / unittest / windows / windows-job (gh) (trunk failure)
  backends/xnnpack/test/ops/test_conv1d.py::TestConv1d::test_qs8_conv1d_batchnorm_seq
• pull / unittest-editable / windows / windows-job (gh) (trunk failure)
  backends/xnnpack/test/ops/test_conv1d.py::TestConv1d::test_qs8_conv1d_batchnorm_seq

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.