Arch Linux

oysteijo

Member

From: Stavanger

Registered: 2012-12-27

Posts: 19

[SOLVED] New laptop - running WSL - nvidia drivers

Hi!

I got a new laptop where I run Arch Linux through WSL. (I'm familiar with Arch and I'm familiar with WSL, but this is the first time I do Arch on WSL)

So - I try to get the nvidia RTX4070 graphics to run in pytorch, but somethings seems to be wrong.
I do something like this:

sudo pacman -S cuda
sudo pacman -S nvidia-open
sudo pacman -S python-pytorch-opt-cuda

However, when I then try pytorch, I get:

[oystein@laptop ~]$ python
Python 3.14.5 (main, May 10 2026, 18:26:20) [GCC 16.1.1 20260430] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> print(torch.cuda.is_available())
/usr/lib/python3.14/site-packages/torch/cuda/__init__.py:187: UserWarning: CUDA initialization: The NVIDIA driver on your system is too old (found version 12080). Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver. (Triggered internally at /build/python-pytorch/src/pytorch-opt-cuda/c10/cuda/CUDAFunctions.cpp:119.)
  return torch._C._cuda_getDeviceCount() > 0
False

What am I missing?

-Øystein

Last edited by oysteijo (Yesterday 14:32:04)

dasrubbellos

Member

Registered: Yesterday

Posts: 5

Re: [SOLVED] New laptop - running WSL - nvidia drivers

Afaik WSL cuda uses your host drivers, you shouldn't need nvidia-open within WSL. Update your host drivers and get rid of nvidia-open.

Your Windows Cuda is 12.8, Arch installs and probably builds against 13.3 currently.

oysteijo

Member

From: Stavanger

Registered: 2012-12-27

Posts: 19

Re: [SOLVED] New laptop - running WSL - nvidia drivers

Thanks. That explains... let me see if I can "downgrade" my pytorch. Or upgrade my host driver.

Update: I managed to upgrade the host driver, and now I get this:

[oystein@laptop ~]$ python
Python 3.14.5 (main, May 10 2026, 18:26:20) [GCC 16.1.1 20260430] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> print(torch.cuda.is_available())
True
>>>

Thank you so much!

Last edited by oysteijo (Yesterday 14:31:30)