[Executorch][llm] Fix flakyness of quantized sdpa test

[kimishpatel]

Stack from ghstack (oldest at bottom):

• [Executorch][LLM] Use caching allocator for runner #15656
• [Executorch] Make module constructors uniform across #15655
• [Executorch] Use temp allocator for allocating scratch memory #15654
• [Executorch] Introduce caching cpu memory allocator #15611
• Reduce allocation overhead in quantized sdpa #15610
• [Executorch] Add multithreading for op_quantize #15609
• [Executorch] Add simd path for op quantize #15608
• [Executorch] parallelize op_choose_qparams #15607
• -> [Executorch][llm] Fix flakyness of quantized sdpa test #15653

Without this fix when running diabled tets it fails with
OMP: Error #131: Thread identifier invalid

Differential Revision: D86447540

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15653

• 📄 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! (5 Unrelated Failures)

As of commit 4d79df5 with merge base 15a0fcd ():

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)
  ##[error]The operation was canceled.
• 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.

[metascroy]

Is there a concern with users choose a thread count that doesn't work?

[kimishpatel]

I think this is specifically related to OMP which I am definitely a bit surprised why it occurs here. This hasnt been observed on-device