CUDA type error building on Windows 10

Hello. I'm trying to install detectron2 on my windows 10. I followed some guides. But it seems this error is brand new.

Instructions To Reproduce the Issue:

Installing by pip install -e . successes. But when running demo.py, it fails with Not compiled with GPU support I think pip install cannot detect my CUDA (even I set PATH settings). So I tried to install with setup.py

Tried to install with setup.py because it detects CUDA OK.


(conda env) python setup.py install
~~~ long warnings ~~~

D:/Projects/detectron2/detectron2/layers/csrc/ROIAlign/ROIAlign_cuda.cu(337): error: no instance of function template "at::cuda::ATenCeilDiv" matches the argument list argument types are: (long long, long)

D:/Projects/detectron2/detectron2/layers/csrc/ROIAlign/ROIAlign_cuda.cu(337): error: no instance of overloaded function "std::min" matches the argument list argument types are: (, long)

D:/Projects/detectron2/detectron2/layers/csrc/ROIAlign/ROIAlign_cuda.cu(394): error: no instance of function template "at::cuda::ATenCeilDiv" matches the argument list argument types are: (int64_t, long)

D:/Projects/detectron2/detectron2/layers/csrc/ROIAlign/ROIAlign_cuda.cu(394): error: no instance of overloaded function "std::min" matches the argument list argument types are: (, long)

4 errors detected in the compilation of "C:/Users/YOONGI~1/AppData/Local/Temp/tmpxft_00001a10_00000000-10_ROIAlign_cuda.cpp1.ii". error: command 'C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\nvcc.exe' failed with exit status 1

But fails with above error.

## Environment:
pytorch 1.4.0 on Anaconda installation
CUDA 10.1 with CuDNN 7.6.5
Visual Studio 2019
Windows 10

sys.platform win32 Python 3.7.6 (default, Jan 8 2020, 20:23:39) [MSC v.1916 64 bit (AMD64)] numpy 1.18.1 detectron2 0.1 @D:\Projects\detectron2\detectron2 detectron2 compiler MSVC 191627030 detectron2 CUDA compiler not available detectron2 arch flags D:\Projects\detectron2\detectron2_C.cp37-win_amd64.pyd; cannot find cuobjdump DETECTRON2_ENV_MODULE PyTorch 1.4.0 @C:\conda\envs\ml\lib\site-packages\torch PyTorch debug build False CUDA available True GPU 0 GeForce GTX 1070 CUDA_HOME C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1 NVCC Not Available Pillow 7.0.0 torchvision 0.5.0 @C:\conda\envs\ml\lib\site-packages\torchvision torchvision arch flags C:\conda\envs\ml\lib\site-packages\torchvision_C.pyd; cannot find cuobjdump cv2 4.1.2

PyTorch built with:

MSVC 191125547
Intel(R) Math Kernel Library Version 2019.0.4 Product Build 20190411 for Intel(R) 64 architecture applications
OpenMP 200203
CUDA Runtime 10.1
NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_37,code=compute_37
CuDNN 7.5.1
Magma 2.5.1
Build settings: BLAS=MKL, BUILD_NAMEDTENSOR=OFF, BUILD_TYPE=Release, CXX_FLAGS=/DWIN32 /D_WINDOWS /GR /w /EHa /MP /bigobj -openmp, DISABLE_NUMA=1, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=OFF, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=OFF, USE_OPENMP=ON, USE_STATIC_DISPATCH=OFF,

facebookresearch / detectron2

CUDA type error building on Windows 10 #848

Instructions To Reproduce the Issue: