Failed to build Pytorch XPU on Windows Server

[DDEle]

🐛 Describe the bug

I got the following error message when build pytorch with xpu-ops, both on BaseKit 2024.2.1 and BaseKit 2025.0.

[7431/7561] Building SYCL device link file caffe2/aten_xpu/src/CMakeFiles/torch_xpu_ops_sycl_kernels.dir/torch_xpu_ops_sycl_kernels_sycl_device_obj.obj
FAILED: caffe2/aten_xpu/src/CMakeFiles/torch_xpu_ops_sycl_kernels.dir/torch_xpu_ops_sycl_kernels_sycl_device_obj.obj D:/dingyi/pytorch/build/caffe2/aten_xpu/src/CMakeFiles/torch_xpu_ops_sycl_kernels.dir/torch_xpu_ops_sycl_kernels_sycl_device_obj.obj 
caffe2\aten_xpu\src\CMakeFiles\torch_xpu_ops_sycl_kernels_sycl_device_obj.obj-a7899af.bat e9da7c94967a2b23
Intel(R) oneAPI DPC++/C++ Compiler for applications running on Intel(R) 64, Version 2025.0.0 Build 20241008
Copyright (C) 1985-2024 Intel Corporation. All rights reserved.



C:\Program Files (x86)\Intel\oneAPI\compiler\2025.0\bin\compiler\clang-offload-bundler: error: 'D:/dingyi/pytorch/build/caff2/aten_xpu/src/CMakeFiles/torch_xpu_ops_sycl_kernels.dir/ATen/native/xpu/sycl/./torch_xpu_ops_sycl_kernels_generated_LerpKernels.cpp.obj': no such file or directory

icx: error: clang-offload-bundler command failed with exit code 1 (use -v to see invocation)

Batch file failed at line 3 with errorcode 1

Notice that it does have the .cpp.obj file on the disk, and the command inside torch_xpu_ops_sycl_kernels_sycl_device_obj.obj-a7899af.bat can be executed successfully when pasted to a powshell prompt (but give the same error when pasted to a CMD promot).

Step-by-step to reproduce:

C:\ProgramData\miniforge3\Scripts\activate
conda activate yi
"C:\Program Files\Microsoft Visual Studio\2022\Professional\VC\Auxiliary\Build\vcvars64.bat"
set XPU_BUNDLE_ROOT=%ProgramFiles(x86)%\Intel\oneAPI
call "%XPU_BUNDLE_ROOT%\compiler\latest\env\vars.bat"
call "%XPU_BUNDLE_ROOT%\ocloc\latest\env\vars.bat"
set CMAKE_SHARED_LINKER_FLAGS=/FORCE:MULTIPLE
set CMAKE_MODULE_LINKER_FLAGS=/FORCE:MULTIPLE
set CMAKE_EXE_LINKER_FLAGS=/FORCE:MULTIPLE
set USE_CUDA=0
python setup.py bdist_wheel > build.log 2>&1

Content of torch_xpu_ops_sycl_kernels_sycl_device_obj.obj-a7899af.bat.txt here (added .txt dur to the limitation of github)

Versions

Torch Sorce version: pytorch/pytorch@7851460

Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A

OS: Microsoft Windows Server 2016 Datacenter (10.0.14393 64-bit)
GCC version: (MinGW-W64 x86_64-msvcrt-posix-seh, built by Brecht Sanders) 13.1.0
Clang version: Could not collect
CMake version: version 3.30.1
Libc version: N/A

Python version: 3.9.19 | packaged by conda-forge | (main, Mar 20 2024, 12:38:46) [MSC v.1929 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-10-10.0.14393-SP0
Is CUDA available: N/A
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A

CPU:
----------------------
Name: Intel(R) Xeon(R) Platinum 8260L CPU @ 2.30GHz
Manufacturer: GenuineIntel
Family: 179
Architecture: 9
ProcessorType: 3
DeviceID: CPU0
CurrentClockSpeed: 2295
MaxClockSpeed: 2295
L2CacheSize: 24576
L2CacheSpeed: None
Revision: 21765
----------------------
Name: Intel(R) Xeon(R) Platinum 8260L CPU @ 2.30GHz
Manufacturer: GenuineIntel
Family: 179
Architecture: 9
ProcessorType: 3
DeviceID: CPU1
CurrentClockSpeed: 2295
MaxClockSpeed: 2295
L2CacheSize: 24576
L2CacheSpeed: None
Revision: 21765

Versions of relevant libraries:
[pip3] numpy==1.26.4
[conda] numpy                     1.26.4                   pypi_0    pypi