Recent Update Broke Cuda

Hi everyone,

I’ve been using cuda 10.1 without problems for about a year, by ignoring cuda upgrades in pacman.conf. However, my latest pacman -Syu has now broken cuda and I’m not sure what is the conflict or how to fix it.

Specifically, I am able to compile a cuda program still, but when I try running a cuda program it throws an unknown error on the first cuda library call. I verified it does this not just with my code, but also with the packaged cuda samples.

Anyone know what the problem is? Has something changed regarding the nvidia drivers? What is the most pragmatic and easy way to solve this and prevent it from happening again? Thank you very much if you can help, this is turning out to be a major pain.

Some relevant info:

$cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module  430.64  Sun Oct 27 11:26:12 UTC 2019
GCC version:  gcc-Version 10.1.0 (GCC) 

$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243

Thu Aug 27 00:15:46 2020       
| NVIDIA-SMI 430.64       Driver Version: 430.64       CUDA Version: 10.1     |
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|   0  GeForce GTX 970     Off  | 00000000:01:00.0  On |                  N/A |
|  0%   46C    P5    18W / 200W |    201MiB /  4034MiB |      0%      Default |

Hello and welcome,

CUDA is now at version 11.0.2-1
Remove cuda you have installed, remove the 430 nvidia drivers (make sure you backup the /etc/X11/mhwd.d/nvidia.conf if you have a custom one), install nvidia 450 drivers, then install the new cuda, restore your configuration. Reboot.

Everything should be back to normal after that.

1 Like

Ok, I’ll give that a try. Thanks.

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.