[windows] ut hang

### 🐛 Describe the bug

Hang during execution of following uts:

- pytorch/third_party/torch-xpu-ops/test/xpu/run_test_with_only.py
- log: [ut_only.log](https://github.com/user-attachments/files/23957236/ut_only.log)
- pytorch/third_party/torch-xpu-ops/test/xpu/run_test_with_skip.py
- pytorch/third_party/torch-xpu-ops/test/xpu/extended/run_test_with_skip.py
- log: [ut_extended.log](https://github.com/user-attachments/files/23957235/ut_extended.log)

### Versions

Click to expand

DLE: 2025.3.1.29
PyTorch version: 2.10.0a0+git7716da9
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: Microsoft Windows 11 Pro (10.0.26100 64-bit)
GCC version: Could not collect
Clang version: Could not collect
CMake version: version 3.31.6
Libc version: N/A

Python version: 3.12.12 | packaged by conda-forge | (main, Oct 22 2025, 23:13:34) [MSC v.1944 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-11-10.0.26100-SP0
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
Is XPU available: True
XPU used to build PyTorch: 20250301
Intel GPU driver version:
* 32.0.101.8250 (20251107000000.******+***)
Intel GPU models onboard:
* Intel(R) Arc(TM) B580 Graphics
Intel GPU models detected:
* [0] _XpuDeviceProperties(name='Intel(R) Arc(TM) B580 Graphics', platform_name='Intel(R) oneAPI Unified Runtime over Level-Zero V2', type='gpu', device_id=0xE20B, uuid=86800be2-0000-0000-0400-000000000000, driver_version='1.13.35227', total_memory=11869MB, max_compute_units=160, gpu_eu_count=160, gpu_subslice_count=20, max_work_group_size=1024, max_num_sub_groups=64, sub_group_sizes=[16 32], has_fp16=1, has_fp64=1, has_atomic64=1)
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Name: Intel(R) Core(TM) Ultra 5 245K
Manufacturer: GenuineIntel
Family: 773
Architecture: 9
ProcessorType: 3
DeviceID: CPU0
CurrentClockSpeed: 3071
MaxClockSpeed: 4200
L2CacheSize: 26624
L2CacheSpeed: None
Revision: None

Versions of relevant libraries:
[pip3] intel-openmp==2025.3.1
[pip3] mkl-include==2025.3.0
[pip3] mkl-static==2025.3.0
[pip3] mypy==1.19.0
[pip3] mypy_extensions==1.1.0
[pip3] numpy==1.26.2
[pip3] onemkl-license==2025.3.0
[pip3] onnx==1.19.1
[pip3] onnx-ir==0.1.12
[pip3] onnxscript==0.5.4
[pip3] optree==0.13.0
[pip3] pytorch-labs-segment-anything-fast==0.2
[pip3] pytorch-triton-xpu==3.5.0+git1b0418a9
[pip3] tbb==2022.3.0
[pip3] tbb-devel==2022.3.0
[pip3] tcmlib==1.4.1
[pip3] torch==2.10.0a0+git7716da9
[pip3] torch_geometric==2.4.0
[pip3] torchao==0.14.1
[pip3] torchaudio==2.10.0a0+ffaeefe
[pip3] torchbench==0.1
[pip3] torchmetrics==1.8.2
[pip3] torchmultimodal==0.1.0b0
[pip3] torchrec-nightly==2022.4.26
[pip3] torchtext==0.17.0a0+a5e6106
[pip3] torchvision==0.25.0a0+6b56de1
[pip3] torchx-nightly==2025.12.4
[conda] intel-openmp 2025.3.1 pypi_0 pypi
[conda] mkl-include 2025.3.0 pypi_0 pypi
[conda] mkl-static 2025.3.0 pypi_0 pypi
[conda] numpy 1.26.2 pypi_0 pypi
[conda] onemkl-license 2025.3.0 pypi_0 pypi
[conda] optree 0.13.0 pypi_0 pypi
[conda] pytorch-labs-segment-anything-fast 0.2 pypi_0 pypi
[conda] pytorch-triton-xpu 3.5.0+git1b0418a9 pypi_0 pypi
[conda] tbb 2022.3.0 pypi_0 pypi
[conda] tbb-devel 2022.3.0 pypi_0 pypi
[conda] tcmlib 1.4.1 pypi_0 pypi
[conda] torch 2.10.0a0+git7716da9 pypi_0 pypi
[conda] torch-geometric 2.4.0 pypi_0 pypi
[conda] torchao 0.14.1 pypi_0 pypi
[conda] torchaudio 2.10.0a0+ffaeefe pypi_0 pypi
[conda] torchbench 0.1 pypi_0 pypi
[conda] torchmetrics 1.8.2 pypi_0 pypi
[conda] torchmultimodal 0.1.0b0 pypi_0 pypi
[conda] torchrec-nightly 2022.4.26 pypi_0 pypi
[conda] torchtext 0.17.0a0+a5e6106 pypi_0 pypi
[conda] torchvision 0.25.0a0+6b56de1 pypi_0 pypi
[conda] torchx-nightly 2025.12.4 pypi_0 pypi

Start by reproducing the hang with pytorch/third_party/torch-xpu-ops/test/xpu/run_test_with_only.py, run_test_with_skip.py, and test/xpu/extended/run_test_with_skip.py on Windows 11. Compare the behavior with ut_only.log and ut_extended.log, then trace the first test or command that stops progressing. Done means the listed test runs complete without hanging.

Written by the indexing model from the issue text.