Installing NVIDIA drivers on Linux always means the same checks: which GPU you have, whether it needs the open or the proprietary kernel module, whether the headers match the running kernel, and what to do about Secure Boot. SysLinuxOS NVIDIA Setup automates all of it on SysLinuxOS 13 and Debian 13 amd64: a Bash script with a YAD graphical interface and a terminal alternative. Source code: github.com/fconidi/syslinuxos-nvidia-setup.
1. What it does
It detects NVIDIA GPUs (including hybrid laptops), picks the right module from NVIDIA’s official JSON database and installs the driver through APT using signed repositories. CUDA is optional. It asks for administrative privileges only at install time, never restarts the computer and never touches Secure Boot.
2. How the driver is chosen
Modern GPUs that support the open kernel module use NVIDIA’s Debian 13 repository with nvidia-open and nvidia-kernel-open-dkms. GPUs that need the proprietary module use Debian’s nvidia-driver and nvidia-kernel-dkms, with an extra nvidia-detect check. Unknown GPUs and legacy branches older than 580 are reported without forcing anything. The two channels are never mixed, CUDA included.
Headers must match the running kernel: the script does not install or select another kernel. Debian’s 550 drivers are blocked on kernel 6.19 and newer, including SysLinuxOS kernel 7.0, so you need to boot a compatible kernel before retrying.
3. Install from the APT repository
The package is distributed through the signed SysLinuxOS-Tools repository. After setting it up with the linked instructions:
sudo apt update sudo apt install --reinstall yad syslinuxos-nvidia-setup syslinuxos-nvidia-setup --check --cli
The repository also works on Debian 13 amd64: the tirreno suite only identifies this extra repository, while Debian’s own repositories stay on trixie. --check shows what would be done without installing anything, without root and without changing the system.
4. Graphical use
In the MATE menu open Applications → SysLinuxOS-Tools → SysLinuxOS NVIDIA Setup. Installing the package does not start any driver installation. In the window choose Driver only or Driver + CUDA, then authenticate: downloads, installation and DKMS compilation continue automatically. The interface supports English, Italian, Spanish, German and French and follows the session language.
5. Terminal use
./syslinuxos-nvidia-setup.sh --cli # asks whether to install CUDA sudo ./syslinuxos-nvidia-setup.sh --cli --no-cuda # no prompts sudo ./syslinuxos-nvidia-setup.sh --cli --cuda # includes CUDA
The portable version needs YAD and pkexec for the GUI: sudo apt install yad pkexec.
6. Secure Boot and verification after restarting
With Secure Boot enabled, if the module signing key is not enrolled the installation ends with exit code 20 and MOK instructions. With the standard DKMS configuration:
sudo mokutil --import /var/lib/dkms/mok.pub
Set a temporary password and restart. In the firmware screen choose Enroll MOK → Continue → Yes and enter the password. After restarting:
nvidia-smi lsmod | grep nvidia dkms status nvcc --version # only if CUDA was selected; open a new terminal
The CUDA Version field in nvidia-smi describes driver compatibility, not an installed toolkit: use nvcc for that.
7. Logs and removal
The administrative log is /var/log/syslinuxos-nvidia-XXXXXXXX.log. APT, DKMS and initramfs errors stop the operation, but the installation can be partial: there is no automatic rollback. To remove only the utility:
sudo apt purge syslinuxos-nvidia-setup
Drivers, CUDA and APT configuration stay installed. If graphics problems occur, boot a working kernel from the GRUB menu or use a text console, read the log and plan the package removal through APT.
Autoinstall NVIDIA Drivers and CUDA on SysLinuxOS 13 and Debian 13