CUDA Toolkit is not detected on Linux

[YAFU]

trafficstars

Hi. Using Version 0.1 - alpha addon. The addon informs me: Python Version: 3.9.2 CUDA Toolkit not found

I have manually installed CUDA Toolkit 11.1.105 from .run file on Kubuntu 20.04 (Linux): $ nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Mon_Oct_12_20:09:46_PDT_2020 Cuda compilation tools, release 11.1, V11.1.105 Build cuda_11.1.TC455_06.29190527_0

CUDA Toolkit is working on my system, for example to build CUDA Kernels for Cycles/Blender.

[YAFU]

Just in case the following information: When you install the CUDA Toolkit from the .run file, you need to set the PATH variables in ~/.bashrc for it to be detected correctly, as explained in the official documentation(9. Post-installation Actions): https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions

I have done that and as I mentioned before, CUDA is correctly installed on my system. GTX 960 4GB, 460.84 driver

[pKrime]

Hi,

will it work if you launch blender from a bash session? If this is the case, then we have to add CUDA to the PATH in the blender launcher as well.

I should add a manual fallback anyway, thanks for your detailed info

[YAFU]

Oh that's right. CUDA is detected if I launch Blender from the terminal. Here the lines that I had used for my version of CUDA toolkit:

export PATH=/usr/local/cuda-11.1/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

[WACOMalt]

How can I do this on windows? Cuda 11.1 is also not detected for me. I launch via Blender Launcher.

[pKrime]

How can I do this on windows? Cuda 11.1 is also not detected for me. I launch via Blender Launcher.

Hi, what happens if you open a command prompt and type

nvcc --version

When you press return, you should see the Cuda info

nvcc: NVIDIA (R) Cuda compiler driver
...

if that happens, brignet is at fault for not being able to replicate this result. If it doesn't, Cuda should be added to the path. As far as I know, that should be standard in windows, you can try to reinstall cuda and restart windows, or you can add the path manually to the windows settings.