What this does

Installs the NVIDIA driver and CUDA toolkit so that GPU-accelerated workloads run on the hardware. After completion, nvidia-smi reports the GPU and nvcc --version shows the toolkit version.

Steps

Update the package index.
```
sudo apt update
```
Expected output: Package lists updated.
Install the NVIDIA driver meta-package.
```
sudo apt install -y ubuntu-drivers-common
sudo ubuntu-drivers autoinstall
```
Expected output: Driver packages installed.
Reboot the system. The driver loads as a kernel module.
```
sudo reboot
```
Verify the driver installation.
```
nvidia-smi
```
Expected output: Table listing GPU, driver version, and memory usage.
Install the CUDA toolkit (optional for nvcc).
```
sudo apt install -y cuda
```
Expected output: CUDA toolkit installed.

Verification

nvidia-smi --query-gpu=name,driver_version,memory.total --format=csv
# Expected: GPU name, driver version, total memory in MB

Common failures

No GPU detected by lspci — GPU not seated properly. Physically reseat the card.
Secure Boot blocks the driver — Disable Secure Boot in BIOS or sign modules with MOK enrollment.
Driver version mismatch — Run ubuntu-drivers list and manually install a compatible version.
Reboot hangs at Plymouth — Add nvidia-drm.modeset=1 to GRUB kernel parameters.
CUDA installed but nvcc not found — Source /usr/local/cuda/bin/nvcc or add it to the shell profile.

Operator checkpoint

Before treating this as solved, write down the local runtime, model or package version, hardware/backend if relevant, and the verification output. This keeps the guide useful as a Will-It-Run style decision instead of a one-off command transcript.

How to install CUDA drivers on Ubuntu

What this does

Steps

Verification

Common failures

Operator checkpoint

Operator checkpoint

Related guides