Closed YouthImagination closed 9 months ago
I think I install cuda visual studio integration 11.6 correctly. Just in case, I re-install cuda again, and install all component.
But I get same problem CUDA not found
Try setting CMAKE_CUDA_ARCHITECTURES
to just 75
, as per https://github.com/mmp/pbrt-v4/issues/374#issuecomment-1652419182
Thanks for reply! I try to set CMAKE_CUDA_ARCHITECTURES
just to 75
and change cuda version to 12.1. But I got same error, it seems not work for me
Found CUDA: 12.1.105
The CUDA compiler identification is unknown
CMake Error at D:/software/CMake/share/cmake-3.23/Modules/CMakeDetermineCUDACompiler.cmake:654 (message):
The CMAKE_CUDA_ARCHITECTURES:
75
do not all work with this compiler. Try:
instead.
Call Stack (most recent call first):
CMakeLists.txt:191 (enable_language)
Configuring incomplete, errors occurred!
See also "D:/Codes/pbrt-v4/build/CMakeFiles/CMakeOutput.log".
See also "D:/Codes/pbrt-v4/build/CMakeFiles/CMakeError.log".
My Cmake
version is 3.23.0-rc5
. I set CMAKE_CUDA_COMPILER
to C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v12.1/bin/nvcc.exe
. I am not sure if it is correnct.
When I build with cmake-gui or cmd. cmake can't find cuda compiler Automatically. And cmake give output log
I alredy add cuda path to enviroment, set
CMAKE_CUDA_COMPILER
withC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\bin\nvcc.exe
and setPBRT_OPTIX7_PATH
value withD:/software/NVIDIA Corporation/OptiX SDK 7.3.0
Then cmake give logMy display card is
Nvidia GeForce GTX1660Ti
so I specifyCMAKE_CUDA_ARCHITECTURES
tosm_75
. cmake error isI try another arch, but it didn't work. How could I deal with this.