Nvidia Fedora Notes

Table of Contents

1. Driver Installation

1. Driver Installation

1.1. Prerequisites

1.1.1. Install the gcc Compiler

Install gcc:
```
sudo dnf install gcc -y
```

Install gcc Development Tools

sudo dnf groupinstall "Development Tools" -y

1.1.2. Install Libraries

Install libs

sudo dnf install -y pkgconf-pkg-config libglvnd-devel

1.1.3. Add NVIDIA Repo

Add NVIDIA Container Toolkit repo

sudo wget -O /etc/yum.repos.d/nvidia-container-toolkit.repo \
  https://nvidia.github.io/libnvidia-container/stable/rpm/nvidia-container-toolkit.repo

Update package meta data

sudo dnf clean expire-cache

sudo dnf -y makecache

1.1.4. Confirm Secure Boot is not enabled

Run

mokutil --sb-state

Sample Result

SecureBoot disabled
Platform is in Setup Mode

1.1.5. Identify Graphics Card

Run

lspci | grep -E "VGA|3D"

Sample Result

2a:00.0 VGA compatible controller: NVIDIA Corporation GA102 [GeForce RTX 3090] (rev a1)

1.2. Fedora `Nvidia Driver` Install

Update System
```
sudo dnf -y update
```
Download the Nvidia Driver for Linux

You’ll need to download a Linux driver from the actual Nvidia drivers downloads page.
Example File to download
```
NVIDIA-Linux-x86_64-570.153.02.run
```

Make the file executable:

Example

chmod +x NVIDIA-Linux-x86_64-570.153.02.run

Run the executable:

Example

sudo ./NVIDIA-Linux-x86_64-570.153.02.run

For this interactive install, use these options:

Kernal module type	NVIDIA Proprietary
Create modprobe entries	Yes
Not able to perform sanity checks	Continue Installation
Install 32 bit libraries	Yes
Register kernal modules with DKMS	Yes
Would you like to rebuild the initramfs	Rebuild initramfs
Run nvidia-xconfig to configure X	Yes

Reboot after the installation
```
sudo reboot
```

1.3. Confirm Installation

After the previous reboot, confirm the installation:

nvidia-smi

Show sample output

Sun May 25 10:52:43 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.153.02             Driver Version: 570.153.02     CUDA Version: 12.8     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3090        Off |   00000000:2A:00.0 Off |                  N/A |
|  0%   59C    P0             72W /  350W |       0MiB /  24576MiB |      4%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

1.4. Install the `Nvidia Tookit`

Install the Nvidia Toolkit

sudo dnf -y install nvidia-container-toolkit

Generate the CDI specification:

sudo nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml

Confirm CDI config.

sudo nvidia-ctk cdi list info

Sample result

INFO[0000] Found 3 CDI devices
nvidia.com/gpu=0
nvidia.com/gpu=GPU-12345678-12345-12345-1234-123456789abc
nvidia.com/gpu=all

Restart Container daemon
Podman
```
sudo systemctl restart podman.socket
```

1.5. Create Service to recreate `Nvidia CDI`

This service is a bandaid and shouldn’t be required.
It will regenerate the CDI specification automatically upon booting the system.

Create the service file

sudo vi /etc/systemd/system/bandaid-nvidia-cdi.service

Add these contents

[Unit]
Description=This is a bandaid for generating the NVIDIA CDI Configuration upon boot
After=network.target

[Service]
ExecStart=nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml
User=root
Group=root
RemainAfterExit=yes

[Install]
WantedBy=multi-user.target

Reload the systemd configuration:
```
sudo systemctl daemon-reload
```

Start the service and enable it to start at boot:

sudo systemctl enable --now bandaid-nvidia-cdi.service