Cuda not found on windows 10

[Fqlox]

Hi, I wanted to build the project on windows 10 but got some issue when preparing the build.

I got :

-- Building for: Visual Studio 16 2019
-- Selecting Windows SDK version 10.0.19041.0 to target Windows 10.0.19044.
-- The C compiler identification is MSVC 19.29.30136.0
-- The CXX compiler identification is MSVC 19.29.30136.0
-- The CUDA compiler identification is unknown
CMake Error at G:/Program Files/CMake/share/cmake-3.23/Modules/CMakeDetermineCUDACompiler.cmake:633 (message):
  Failed to detect a default CUDA architecture.



  Compiler output:

Call Stack (most recent call first):
  CMakeLists.txt:11 (PROJECT)

I did reinstall CUDA and update it to the path to the latest version with the visual studio integration.
And add the files in the VS directory as expected.

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_Mar__8_18:36:24_Pacific_Standard_Time_2022
Cuda compilation tools, release 11.6, V11.6.124
Build cuda_11.6.r11.6/compiler.31057947_0

Thanks in advance

[kdand35]

Same problems for me, waiting for the solutions.

[Ardibid]

Same issue here! It would be great if you could help with this.

[bdweix]

Same issue here - working through some troubleshooting now. I’ll update if I figure it out!

[spacemishka]

Same here....why is this always such a PITA?!

[greenbrettmichael]

are you running from developer command prompt or developer powershell?

[Fqlox]

are you running from developer command prompt or developer powershell?

I tested on both, also with and without "run as administrator"

[greenbrettmichael]

Thank you for your quick reply while I am still looking at this!
I am not affiliated with the project but I just got this up and running on Windows.

A couple of diagnostic questions that might be helpful
it says that you are using the VS 2019 generator, do you have VS 2022 installed as well? that could be a potential source of errors if the VS integration is installed to one particular installation

do you see the path similar to
C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.6/bin
in your system path? also have you reset your computer since the CUDA tk was installed ? powershell can be quite sticky about env vars depending on host terminal

[Fqlox]

A couple of diagnostic questions that might be helpful it says that you are using the VS 2019 generator, do you have VS 2022 installed as well? that could be a potential source of errors if the VS integration is installed to one particular installation

I got also VS 2017 installed, It got installed with Unity.

do you see the path similar to C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.6/bin in your system path? also have you reset your computer since the CUDA tk was installed ? powershell can be quite sticky about env vars depending on host terminal

I got CUDA installed to my path, Not on C: but on HDD and it works ("nvcc" prompt works, and conda env with pytorch/tensorflow load CUDA).

[kdand35]

This is my current error, I am using VS 2019 and cuda 11.5, still cannot fix it so far.......
C:\instant-ngp\instant-ngp-master>cmake . -B build
-- The CUDA compiler identification is unknown
-- Check for working CUDA compiler: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.5/bin/nvcc.exe
CMake Error in C:/instant-ngp/instant-ngp-master/build/CMakeFiles/CMakeTmp/CMakeLists.txt:
CUDA_ARCHITECTURES is empty for target "cmTC_38f99".

CMake Error in C:/instant-ngp/instant-ngp-master/build/CMakeFiles/CMakeTmp/CMakeLists.txt:
CUDA_ARCHITECTURES is empty for target "cmTC_38f99".

CMake Error at C:/Program Files/CMake/share/cmake-3.21/Modules/CMakeTestCUDACompiler.cmake:39 (try_compile):
Failed to generate test project build system.
Call Stack (most recent call first):
CMakeLists.txt:11 (PROJECT)

-- Configuring incomplete, errors occurred!
See also "C:/instant-ngp/instant-ngp-master/build/CMakeFiles/CMakeOutput.log".
See also "C:/instant-ngp/instant-ngp-master/build/CMakeFiles/CMakeError.log".

[greenbrettmichael]

@Fqlox the reply by @kdand35 made me reread top post
the issue is not "Cuda not found" but "Failed to detect a default CUDA architecture"

you should be able to use CUDA and CMAKE mailing lists to further debug this issue

re: @Fqlox cmake is not populating CUDA_ARCHITECTURES for your build

@kdand35 this is covered in the FAQ with this answer Windows: the Visual Studio CUDA integration was not installed correctly. Follow mitsuba-renderer/mitsuba2#103 (comment) to fix the problem without re-installing CUDA. (#18)

[Fqlox]

@Fqlox the reply by @kdand35 made me reread top post the issue is not "Cuda not found" but "Failed to detect a default CUDA architecture"

you should be able to use CUDA and CMAKE mailing lists to further debug this issue

re: @Fqlox cmake is not populating CUDA_ARCHITECTURES for your build

@kdand35 this is covered in the FAQ with this answer Windows: the Visual Studio CUDA integration was not installed correctly. Follow mitsuba-renderer/mitsuba2#103 (comment) to fix the problem without re-installing CUDA. (#18)

I did this, and still getting the same issue compiling

[kdand35]

Hi, @Fqlox !
I complete the compilation yesterday! Some tips might be helpful for you.

Previously, I only downloaded the zip file of this project and compile it on my computer. To pass the above problems, I just deleted the whole file I downloaded and rebuilt it again. The reason might be I tried to use several versions of Cmake to compile and it made some remains in my build/CMakeFiles path. I am not sure if it is the true reason, it passed the above problems, but I met something new.......

Yesterday, I used Git to clone and compile, and it works. Also, I found a good video to show how to install it and very helpful for the installation. Please also refer: https://www.youtube.com/watch?v=kq9xlvz73Rg&t=563s

I am not sure if there are some differences between these two ways, but really recommend you use Git if you don't use it currently.

Good luck and Hope it helps!

[Fqlox]

I did installed via git, since the --recursive args allows to download dependencies. I did delete and redownload the repo, uninstalled VS 2019 and installed it again, swapped the CUDA File VS need (from C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\extras\visual_studio_integration\MSBuildExtensions to C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\BuildCustomizations ).

I did look the tutorial and did exaclty the same stuff as mentionned : still got the issue while preparing the compilation.

[Fqlox]

I managed to correct this issue by fully uninstalling Visual Studio and CUDA and reinstalling both on the system drive without changing the installation path.