python - nvcc 致命：无法打开输出文件

Question

我使用 pytorch 在 Windows 上进行对象检测算法：

当我从 https://github.com/lufficc/SSD 存储库运行脚本ext/build.py build_ext develop或从 https://github.com/potterhsu/easy-faster-rcnn.pytorch存储库运行时python support/setup.py develop

import glob
import os

import torch
from setuptools import setup
from torch.utils.cpp_extension import CUDA_HOME
from torch.utils.cpp_extension import CppExtension
from torch.utils.cpp_extension import CUDAExtension

requirements = ["torch", "torchvision"]


def get_extensions():
    this_dir = os.path.dirname(os.path.abspath(__file__))
    extensions_dir = os.path.join(this_dir, "src")

    main_file = glob.glob(os.path.join(extensions_dir, "*.cpp"))
    source_cpu = glob.glob(os.path.join(extensions_dir, "cpu", "*.cpp"))
    source_cuda = glob.glob(os.path.join(extensions_dir, "cuda", "*.cu"))

    sources = main_file + source_cpu
    extension = CppExtension

    extra_compile_args = {"cxx": []}
    define_macros = []

    if torch.cuda.is_available() and CUDA_HOME is not None:
        extension = CUDAExtension
        sources += source_cuda
        define_macros += [("WITH_CUDA", None)]
        extra_compile_args["nvcc"] = [
            "-DCUDA_HAS_FP16=1",
            "-D__CUDA_NO_HALF_OPERATORS__",
            "-D__CUDA_NO_HALF_CONVERSIONS__",
            "-D__CUDA_NO_HALF2_OPERATORS__",
        ]

    sources = [os.path.join(extensions_dir, s) for s in sources]

    include_dirs = [extensions_dir]

    ext_modules = [
        extension(
            "support._C",
            sources,
            include_dirs=include_dirs,
            define_macros=define_macros,
            extra_compile_args=extra_compile_args,
        )
    ]

    return ext_modules


setup(
    name="support",
    version="0.1",
    ext_modules=get_extensions(),
    cmdclass={"build_ext": torch.utils.cpp_extension.BuildExtension},
)

我犯了同样的错误：

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc.exe -c C:\Users\along\PycharmProjects\SSD\ext\cuda\nms.cu -o build\temp.win- amd64-3.7\Release\Users\along\PycharmProjects\SSD\ext\cuda\nms.obj -IC:\Users\along\PycharmProjects\SSD\ext -IC:\Users\along\Anaconda3\lib\site-packages\ torch\include -IC:\Users\along\Anaconda3\lib\site-packages\torch\include\torch\csrc\api\include -IC:\Users\along\Anaconda3\lib\site-packages\torch\include\ TH -IC:\Users\along\Anaconda3\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include" -IC:\Users\along \Anaconda3\include -IC:\Users\along\Anaconda3\include "-IC:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Tools\MSVC\14.16.27023\ATLMFC\include" "-我知道了：\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Tools\MSVC\14.16.27023\include" "-IC:\Program Files (x86)\Windows Kits\10\include\10.0.17763.0\ucrt " "-IC:\Program Files (x86)\Windows Kits\10\include\10.0.17763.0\shared" "-IC:\Program Files (x86)\Windows Kits\10\include\10.0.17763.0\um" " -IC:\Program Files (x86)\Windows Kits\10\include\10.0.17763.0\winrt" "-IC:\Program Files (x86)\Windows Kits\10\include\10.0.17763.0\cppwinrt"-Xcompiler /EHsc -Xcompiler /wd4819 -Xcompiler /MD -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=torch_extension -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch= compute_75,code=sm_75 nvcc 致命：无法打开输出文件“along\AppData\Local\Temp/tmpxft_00005b70_00000000”错误：命令“C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc.exe ' 退出状态为 1 失败\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc.exe' 失败，退出状态为 1\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc.exe' 失败，退出状态为 1

我不知道是什么错误，我的电脑上有一个 GPU 设备

我认为问题出在 Anaconda3\pkgs\pytorch-1.3.1-py3.7_cuda92_cudnn7_0\info\test\test\custom_operator 目录上的 CMakeLists.txt 文件上：

# Basic CMake setup
cmake_minimum_required(VERSION 3.1 FATAL_ERROR)
project(custom_ops)

find_package(Torch REQUIRED)

add_library(custom_ops SHARED op.cpp)

add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/../../third_party/pybind11/ ./pybind11)
pybind11_add_module(custom_class SHARED classes.cpp)
target_link_libraries(custom_class PRIVATE "${TORCH_LIBRARIES}")

target_compile_features(custom_ops PUBLIC cxx_range_for)
target_link_libraries(custom_ops "${TORCH_LIBRARIES}")
target_compile_definitions(custom_ops PRIVATE custom_ops_EXPORTS)

add_executable(test_custom_ops test_custom_ops.cpp)
target_link_libraries(test_custom_ops custom_ops)

标签： pythonpytorchcondanvcc