[Installation]: An error occurred when I tried to build from source in the cuda12.8, pytorch2.10.0, win10 environment

[User-Clb]

Your current environment

os：win10
python：3.12.10
CUDA：12.8
GPU：3060
pytorch：2.10.0

How you are installing vllm

set DISTUTILS_USE_SDK=1
set VLLM_TARGET_DEVICE=cuda
set MAX_JOBS=4

set USE_CUDNN=1
set CUDNN_LIBRARY_PATH=D:\xx..\tmp\cuda\v9.17\lib\13.1\x64
set CUDNN_INCLUDE_PATH=D:\xx..\tmp\cuda\v9.17\include\13.1

set USE_CUDSS=1
set CUDSS_LIBRARY_PATH=D:\xx..\tmp\cuda\v0.7\lib\12
set CUDSS_INCLUDE_PATH=D:\xx..\tmp\cuda\v0.7\include

set USE_CUSPARSELT=1
set CUSPARSELT_INCLUDE_PATH=D:\xx..\tmp\cuda\v0.8\include
set CUSPARSELT_LIBRARY_PATH=D:\xx..\tmp\cuda\v0.8\lib

python use_existing_torch.py
pip install -r requirements/build.txt
pip install -r requirements/windows.txt

set TORCH_CUDA_ARCH_LIST=8.6

pip install . --no-build-isolation -v

【Note: D:/xx/../.venv is the environment created for the local uv】

ninja: build stopped: subcommand failed.
[stderr]
D:\xx...venv\Lib\site-packages\setuptools_scm_integration\version_inference.py:51: UserWarning: version of None already set
warnings.warn(self.message)
CMake Warning at D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Caffe2/public/cuda.cmake:149 (message):
Failed to compute shorthash for libnvrtc.so
Call Stack (most recent call first):
D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Caffe2/Caffe2Config.cmake:86 (include)
D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:68 (find_package)
CMakeLists.txt:90 (find_package)

CMake Warning at D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Caffe2/public/cuda.cmake:332 (message):
pytorch is not compatible with CMAKE_CUDA_ARCHITECTURES and will ignore
its value. Please configure TORCH_CUDA_ARCH_LIST instead.
Call Stack (most recent call first):
D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Caffe2/Caffe2Config.cmake:86 (include)
D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:68 (find_package)
CMakeLists.txt:90 (find_package)

CMake Warning at D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:22 (message):
static library kineto_LIBRARY-NOTFOUND not found.
Call Stack (most recent call first):
D:/xx/../.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:125 (append_torchlib_if_found)
CMakeLists.txt:90 (find_package)

CMake Warning (dev) at cmake/external_projects/flashmla.cmake:52 (find_package):
Policy CMP0148 is not set: The FindPythonInterp and FindPythonLibs modules
are removed. Run "cmake --help-policy CMP0148" for policy details. Use
the cmake_policy command to set the policy and suppress this warning.

Call Stack (most recent call first):
CMakeLists.txt:1179 (include)
This warning is for project developers. Use -Wno-dev to suppress it.

CMake Warning (dev) at D:/xx/../.venv/Lib/site-packages/cmake/data/share/cmake-4.2/Modules/FetchContent.cmake:1963 (message):
Calling FetchContent_Populate(qutlass) is deprecated, call
FetchContent_MakeAvailable(qutlass) instead. Policy CMP0169 can be set to
OLD to allow FetchContent_Populate(qutlass) to be called directly for now,
but the ability to call it with declared details will be removed completely
in a future version.
Call Stack (most recent call first):
cmake/external_projects/qutlass.cmake:27 (FetchContent_Populate)
CMakeLists.txt:1180 (include)
This warning is for project developers. Use -Wno-dev to suppress it.

CMake Warning at .deps/vllm-flash-attn-src/CMakeLists.txt:77 (message):
Pytorch version 2.4.0 expected for CUDA build, saw 2.10.0 instead.

CMake Warning (dev) at cmake/external_projects/vllm_flash_attn.cmake:66 (find_package):
Policy CMP0148 is not set: The FindPythonInterp and FindPythonLibs modules
are removed. Run "cmake --help-policy CMP0148" for policy details. Use
the cmake_policy command to set the policy and suppress this warning.

Call Stack (most recent call first):
CMakeLists.txt:1183 (include)
This warning is for project developers. Use -Wno-dev to suppress it.

CMake Warning:
Manually-specified variables were not used by the project:

nvtx3_dir

Traceback (most recent call last):
File "", line 11, in
File "D:\xx...venv\Lib\site-packages\setuptools\build_meta.py", line 439, in build_wheel
return _build(['bdist_wheel', '--dist-info-dir', str(metadata_directory)])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\xx...venv\Lib\site-packages\setuptools\build_meta.py", line 427, in _build
return self._build_with_temp_dir(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\xx...venv\Lib\site-packages\setuptools\build_meta.py", line 408, in build_with_temp_dir
self.run_setup()
File "D:\xx...venv\Lib\site-packages\setuptools\build_meta.py", line 317, in run_setup
exec(code, locals())
File "", line 1122, in
File "D:\xx...venv\Lib\site-packages\setuptools_init.py", line 117, in setup
return distutils.core.setup(**attrs) # type: ignore[return-value]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\core.py", line 186, in setup
return run_commands(dist)
^^^^^^^^^^^^^^^^^^
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\core.py", line 202, in run_commands
dist.run_commands()
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\dist.py", line 1002, in run_commands
self.run_command(cmd)
File "D:\xx...venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
super().run_command(command)
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\dist.py", line 1021, in run_command
cmd_obj.run()
File "D:\xx...venv\Lib\site-packages\setuptools\command\bdist_wheel.py", line 370, in run
self.run_command("build")
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\cmd.py", line 357, in run_command
self.distribution.run_command(command)
File "D:\xx...venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
super().run_command(command)
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\dist.py", line 1021, in run_command
cmd_obj.run()
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\command\build.py", line 135, in run
self.run_command(cmd_name)
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\cmd.py", line 357, in run_command
self.distribution.run_command(command)
File "D:\xx...venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
super().run_command(command)
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\dist.py", line 1021, in run_command
cmd_obj.run()
File "", line 447, in run
File "D:\xx...venv\Lib\site-packages\setuptools\command\build_ext.py", line 97, in run
_build_ext.run(self)
File "D:\xx...venv\Lib\site-packages\setuptools_distutils\command\build_ext.py", line 368, in run
self.build_extensions()
File "", line 416, in build_extensions
File "D:\xx...venv\Lib\subprocess.py", line 413, in check_call
raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['cmake', '--build', '.', '-j=4', '--target=_moe_C', '--target=cumem_allocator', '--target=triton_kernels', '--target=_vllm_fa2_C', '--target=_C']' returned non-zero exit status 2.

hint: This usually indicates a problem with the package or the build environment.

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[SystemPanic]

Don't see any errors, maybe

CMake Warning at .deps/vllm-flash-attn-src/CMakeLists.txt:77 (message):
Pytorch version 2.4.0 expected for CUDA build, saw 2.10.0 instead.

Try to remove .deps/vllm-flash-attn-src folder and let CMake fetch it again.

If still happens, change set(TORCH_SUPPORTED_VERSION_CUDA "2.4.0") to set(TORCH_SUPPORTED_VERSION_CUDA "2.10.0") in .deps/vllm-flash-attn-src/CMakeLists.txt and PYTORCH_VERSION = "2.4.0" to PYTORCH_VERSION = "2.10.0" in .deps/vllm-flash-attn-src/setup.py

Also, the latest source code of vLLM for Windows is for PyTorch 2.11 nightly dev build. Otherwise, you will face RuntimeError: makeDeviceForHostname(): unsupported gloo device (PyTorch Windows issue fixed on 2.11)

[User-Clb]

Don't see any errors, maybe

CMake Warning at .deps/vllm-flash-attn-src/CMakeLists.txt:77 (message):
Pytorch version 2.4.0 expected for CUDA build, saw 2.10.0 instead.

Try to remove .deps/vllm-flash-attn-src folder and let CMake fetch it again.

If still happens, change set(TORCH_SUPPORTED_VERSION_CUDA "2.4.0") to set(TORCH_SUPPORTED_VERSION_CUDA "2.10.0") in .deps/vllm-flash-attn-src/CMakeLists.txt and PYTORCH_VERSION = "2.4.0" to PYTORCH_VERSION = "2.10.0" in .deps/vllm-flash-attn-src/setup.py

Also, the latest source code of vLLM for Windows is for PyTorch 2.11 nightly dev build. Otherwise, you will face RuntimeError: makeDeviceForHostname(): unsupported gloo device (PyTorch Windows issue fixed on 2.11)

pip install --pre torch==2.11.0.dev20260216+cu128 torchvision==0.26.0.dev20260216+cu128 torchaudio==2.11.0.dev20260216+cu128 --index-url https://download.pytorch.org/whl/nightly/cu128

set DISTUTILS_USE_SDK=1
set VLLM_TARGET_DEVICE=cuda
set MAX_JOBS=8

set USE_CUDNN=1
set CUDNN_LIBRARY_PATH=D:\xx..\v9.17\lib\13.1\x64
set CUDNN_INCLUDE_PATH=D:\xx..\v9.17\include\13.1

set USE_CUDSS=1
set CUDSS_LIBRARY_PATH=D:\xx..\v0.7\lib\12
set CUDSS_INCLUDE_PATH=D\xx..\v0.7\include

set USE_CUSPARSELT=1
set CUSPARSELT_INCLUDE_PATH=D:\xx..\v0.8\include
set CUSPARSELT_LIBRARY_PATH=D:\xx..\v0.8\lib

Compilation failed：
ninja: build stopped: subcommand failed.

  [stderr]
  D:\CC\VLLM\.venv\Lib\site-packages\setuptools_scm\_integration\version_inference.py:51: UserWarning: version of None already set
    warnings.warn(self.message)
  CMake Warning at D:/CC/VLLM/.venv/Lib/site-packages/torch/share/cmake/Caffe2/public/cuda.cmake:332 (message):
    pytorch is not compatible with `CMAKE_CUDA_ARCHITECTURES` and will ignore
    its value.  Please configure `TORCH_CUDA_ARCH_LIST` instead.
  Call Stack (most recent call first):
    D:/CC/VLLM/.venv/Lib/site-packages/torch/share/cmake/Caffe2/Caffe2Config.cmake:86 (include)
    D:/CC/VLLM/.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:68 (find_package)
    CMakeLists.txt:91 (find_package)


  CMake Warning at D:/CC/VLLM/.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:22 (message):
    static library kineto_LIBRARY-NOTFOUND not found.
  Call Stack (most recent call first):
    D:/CC/VLLM/.venv/Lib/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:125 (append_torchlib_if_found)
    CMakeLists.txt:91 (find_package)


  CMake Warning (dev) at cmake/external_projects/flashmla.cmake:52 (find_package):
    Policy CMP0148 is not set: The FindPythonInterp and FindPythonLibs modules
    are removed.  Run "cmake --help-policy CMP0148" for policy details.  Use
    the cmake_policy command to set the policy and suppress this warning.

  Call Stack (most recent call first):
    CMakeLists.txt:1176 (include)
  This warning is for project developers.  Use -Wno-dev to suppress it.

  CMake Warning (dev) at D:/CC/VLLM/.venv/Lib/site-packages/cmake/data/share/cmake-4.2/Modules/FetchContent.cmake:1963 (message):
    Calling FetchContent_Populate(qutlass) is deprecated, call
    FetchContent_MakeAvailable(qutlass) instead.  Policy CMP0169 can be set to
    OLD to allow FetchContent_Populate(qutlass) to be called directly for now,
    but the ability to call it with declared details will be removed completely
    in a future version.
  Call Stack (most recent call first):
    cmake/external_projects/qutlass.cmake:27 (FetchContent_Populate)
    CMakeLists.txt:1177 (include)
  This warning is for project developers.  Use -Wno-dev to suppress it.

  CMake Warning (dev) at cmake/external_projects/vllm_flash_attn.cmake:66 (find_package):
    Policy CMP0148 is not set: The FindPythonInterp and FindPythonLibs modules
    are removed.  Run "cmake --help-policy CMP0148" for policy details.  Use
    the cmake_policy command to set the policy and suppress this warning.

  Call Stack (most recent call first):
    CMakeLists.txt:1180 (include)
  This warning is for project developers.  Use -Wno-dev to suppress it.

  Traceback (most recent call last):
    File "<string>", line 11, in <module>
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\build_meta.py", line 439, in build_wheel
      return _build(['bdist_wheel', '--dist-info-dir', str(metadata_directory)])
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\build_meta.py", line 427, in _build
      return self._build_with_temp_dir(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\build_meta.py", line 408, in _build_with_temp_dir
      self.run_setup()
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\build_meta.py", line 317, in run_setup
      exec(code, locals())
    File "<string>", line 1122, in <module>
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\__init__.py", line 117, in setup
      return distutils.core.setup(**attrs)  # type: ignore[return-value]
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\core.py", line 186, in setup
      return run_commands(dist)
             ^^^^^^^^^^^^^^^^^^
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\core.py", line 202, in run_commands
      dist.run_commands()
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\dist.py", line 1002, in run_commands
      self.run_command(cmd)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
      super().run_command(command)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\dist.py", line 1021, in run_command
      cmd_obj.run()
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\command\bdist_wheel.py", line 370, in run
      self.run_command("build")
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\cmd.py", line 357, in run_command
      self.distribution.run_command(command)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
      super().run_command(command)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\dist.py", line 1021, in run_command
      cmd_obj.run()
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\command\build.py", line 135, in run
      self.run_command(cmd_name)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\cmd.py", line 357, in run_command
      self.distribution.run_command(command)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\dist.py", line 1107, in run_command
      super().run_command(command)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\dist.py", line 1021, in run_command
      cmd_obj.run()
    File "<string>", line 447, in run
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\command\build_ext.py", line 97, in run
      _build_ext.run(self)
    File "D:\CC\VLLM\.venv\Lib\site-packages\setuptools\_distutils\command\build_ext.py", line 368, in run
      self.build_extensions()
    File "<string>", line 416, in build_extensions
    File "D:\CC\SOFT\cude128_comfyui\ComfyUI_windows_portable\py12\Lib\subprocess.py", line 413, in check_call
      raise CalledProcessError(retcode, cmd)
  subprocess.CalledProcessError: Command '['cmake', '--build', '.', '-j=8', '--target=_moe_C', '--target=cumem_allocator', '--target=triton_kernels',
  '--target=_vllm_fa2_C', '--target=_C']' returned non-zero exit status 2.

  hint: This usually indicates a problem with the package or the build environment.

[User-Clb]

DEBUG [3/138] Building CUDA object CMakeFiles_C.dir\csrc\layernorm_quant_kernels.cu.obj
DEBUG FAILED: [code=2] CMakeFiles/_C.dir/csrc/layernorm_quant_kernels.cu.obj
DEBUG "C:\PROGRA~1\NVIDIA GPU Computing Toolkit\CUDA\v12.8\bin\nvcc.exe" -forward-unknown-to-host-compiler -DCUTLASS_ENABLE_DIRECT_CUDA_DRIVER_CALL=1 -DPy_LIMITED_API=3 -DPy_NO_LINK_LIB -DTORCH_EXTENSION_NAME=_C -DUSE_C10D_GLOO -DUSE_DISTRIBUTED -D_C_EXPORTS -ID:\CC\VLLM\vllm-windows\csrc -ID:\VLLM\vllm-windows.deps\cutlass-src\include -ID:\VLLM\vllm-windows.deps\cutlass-src\tools\util\include -isystem D:..\Include -isystem D:\VLLM.venv\Lib\site-packages\torch\include -isystem D:\VLLM.venv\Lib\site-packages\torch\include\torch\csrc\api\include -isystem "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\include" -D_WINDOWS -Xcompiler=" /EHsc" -DONNX_NAMESPACE=onnx_c2 --use-local-env -Xcudafe --diag_suppress=cc_clobber_ignored,--diag_suppress=field_without_dll_interface,--diag_suppress=base_class_has_different_dll_interface,--diag_suppress=dll_interface_conflict_none_assumed,--diag_suppress=dll_interface_conflict_dllexport_assumed,--diag_suppress=bad_friend_decl --Werror cross-execution-space-call --no-host-device-move-forward --expt-relaxed-constexpr --expt-extended-lambda -Xcompiler=" -O2 -Ob1" -DNDEBUG -Xcompiler /MD -std=c++17 -Xcompiler=-MD -Xcompiler=-Zi --expt-relaxed-constexpr -DENABLE_FP8 --threads=1 --disable-warnings -O2 -FS -Xptxas=-O2 -Xcompiler=/O2 -Xcompiler=/FS -Xcompiler=/Z7 -Xcompiler=/Zc:__cplusplus -Xcompiler=/DUSE_CUDA -DENABLE_SCALED_MM_C2X=1 -gencode arch=compute_86,code=sm_86 -MD -MT CMakeFiles_C.dir\csrc\layernorm_quant_kernels.cu.obj -MF CMakeFiles_C.dir\csrc\layernorm_quant_kernels.cu.obj.d -x cu -c D:\VLLM\vllm-windows\csrc\layernorm_quant_kernels.cu -o CMakeFiles_C.dir\csrc\layernorm_quant_kernels.cu.obj -Xcompiler=-FdCMakeFiles_C.dir,-FS
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG layernorm_quant_kernels.cu
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG layernorm_quant_kernels.cu
DEBUG D:\CC\VLLM.venv\Lib\site-packages\torch\include\c10/cuda/CUDACachingAllocator.h(105): error: invalid combination of type specifiers
DEBUG StreamSegmentSize(cudaStream_t s, bool char, size_t sz)
DEBUG ^
DEBUG
DEBUG D:\CC\VLLM.venv\Lib\site-packages\torch\include\c10/cuda/CUDACachingAllocator.h(106): error: type name is not allowed
DEBUG : stream(s), is_small_pool(char), total_size(sz) {}
DEBUG ^
DEBUG
DEBUG 2 errors detected in the compilation of "D:/CC/VLLM/vllm-windows/csrc/layernorm_quant_kernels.cu".

[User-Clb]

DEBUG [1/144] Building CXX object CMakeFiles\cumem_allocator.dir\csrc\cumem_allocator.cpp.obj
DEBUG FAILED: [code=2] CMakeFiles/cumem_allocator.dir/csrc/cumem_allocator.cpp.obj
DEBUG D:...\MICROS1\2022\BUILDT1\VC\Tools\MSVC\1438~1.331\bin\Hostx64\x64\cl.exe /nologo /TP -DPy_LIMITED_API=0x03080000 -DPy_NO_LINK_LIB -DTORCH_EXTENSION_NAME=cumem_allocator -DUSE_C10D_GLOO -DUSE_DISTRIBUTED -Dcumem_allocator_EXPORTS -ID:\VLLM\vllm-windows\csrc -external:ID:..\Include -external:ID:\VLLM.venv\Lib\site-packages\torch\include -external:ID:\VLLM.venv\Lib\site-packages\torch\include\torch\csrc\api\include -external:I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\include" -external:W0 /DWIN32 /D_WINDOWS /EHsc -IC:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v12.8/include/Zc:__cplusplus /O2 /Ob1 /DNDEBUG -std:c++17 -MD -Zi /permissive- /Z7 /EHsc /bigobj /showIncludes /FoCMakeFiles\cumem_allocator.dir\csrc\cumem_allocator.cpp.obj /FdCMakeFiles\cumem_allocator.dir\ /FS -c D:\VLLM\vllm-windows\csrc\cumem_allocator.cpp
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG NVIDIA
DEBUG c1xx: fatal error C1083: �޷���Դ�ļ�: ��Files/NVIDIA��: No such file or directory
DEBUG GPU
DEBUG c1xx: fatal error C1083: �޷���Դ�ļ�: ��GPU��: No such file or directory
DEBUG Computing
DEBUG c1xx: fatal error C1083: �޷���Դ�ļ�: ��Computing��: No such file or directory
DEBUG Zc:__cplusplus
DEBUG c1xx: fatal error C1083: �޷���Դ�ļ�: ��Toolkit/CUDA/v12.8/include/Zc:__cplusplus��: No such file or directory
DEBUG �������ɴ���...
DEBUG [2/144] Building CUDA object CMakeFiles_C.dir\csrc\pos_encoding_kernels.cu.obj
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG pos_encoding_kernels.cu
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG pos_encoding_kernels.cu
DEBUG pos_encoding_kernels.cu
DEBUG cl: ������ warning D9025 :������д��/Zi��(�á�/Z7��)
DEBUG tmpxft_00006094_00000000-7_pos_encoding_kernels.cudafe1.cpp

[SystemPanic]

Now it's crystal clear.

PyTorch CUDACachingAllocator.h has a conflict with the Windows macro small.

Pull the latest changes from vllm-for-windows branch and try to build again.