Why does rpmfusion CUDA install HOWTO say "sudo dnf module disable nvidia-driver" - disable nvidia-driver?

Trying to install CUDA on my ThinkPad is confusing. It appears that the rpmfusion docs want me to disable the NVIDIA driver:

see Howto/CUDA - RPM Fusion

sudo dnf config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/fedora32/x86_64/cuda-fedora32.repo
sudo dnf clean all
sudo dnf module disable nvidia-driver
sudo dnf -y install cuda

Will this turn off my NVIDIA drivers?

Are you installing a stand-alone version of cuda?
I am running the nvidia driver and cuda with both installed from the rpmfusion repo.
To install them both at the same time on a clean install of both f32 and f33 I used

dnf install akmod-nvidia nvidia* xorg-x11-drivers-nvidia* kernel-devel kernel-headers

and everything works both cuda and nvidia related.

No, that dnf module disable command will not turn off the nvidia drivers. What it does is disable the modular version of the nvidia drivers from the repo to avoid installation conflicts and ensure you only get cuda from the newly enabled repo.

I personally have never had a need to install from (or even enable) that cuda repo. I have also helped others that have seen conflicts remove software from there and go only with the rpmfusion version since it has specifically been tailored to work with the nvidia driver.

No, I already installed the NVIDIA drivers from rpmfusion and they are working good. So I assumed I needed to install the CUDA libs and headers separately.

I would suppose that if you are compiling cuda software you need that, but the xorg cuda driver from rpmfusion works for me running boinc with gpu tasks under cuda.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.