How to check that your GPU server can see its NVIDIA GPU?

Updated: 28 September 2026 · Applies to: NVIDIA GPU servers on Ubuntu 24.04 / 26.04 LTS

After your GPU server is delivered, first confirm that the operating system can see the graphics card or cards. Do this before you install any AI software: it tells you in one minute whether the hardware is visible and whether the NVIDIA driver is already installed.

Check the GPU

Log in to the server over SSH as root or as a user with sudo rights.
List the NVIDIA cards that the server sees on its PCI bus:
```
lspci | grep -i nvidia
```
If the command is not found, install it with sudo apt install -y pciutils and run it again. You should get one line per GPU, for example four lines for a server with four A100 cards. GeForce and RTX cards may also list a small audio device next to the GPU; that is normal.
Ask the driver for its view of the cards:
```
nvidia-smi
```
A table appears with the driver version, the GPU model, its memory, temperature and power use. This means the driver works and the GPU is ready.
List every GPU with its number and unique ID:
```
nvidia-smi -L
```
Keep the numbers (0, 1, 2 ...): you will need them to choose a GPU for an application or a container.
Check that the driver's kernel modules are loaded:
```
lsmod | grep nvidia
```

What the results mean

lspci lists the GPU and nvidia-smi prints a table: hardware and driver are fine. Install your software, for example How to install PyTorch with GPU (CUDA) support and test it? or How to install Ollama on Ubuntu and run your first private LLM?.
lspci lists the GPU but nvidia-smi: command not found: the NVIDIA driver is not installed yet. Follow How to install the NVIDIA driver on Ubuntu 24.04 / 26.04 for a GPU server?.
lspci lists the GPU and nvidia-smi says it could not communicate with the driver: the driver is installed but not loaded. Follow How to fix the "NVIDIA-SMI has failed" error (cannot communicate with the NVIDIA driver)?.
lspci shows no NVIDIA device: the server does not see the card at all. Open a support ticket with the output of lspci; this needs a hardware check.

Two things that confuse many people

The CUDA Version in the top right corner of nvidia-smi is the newest CUDA version that the installed driver supports. It does not mean the CUDA Toolkit is installed. The toolkit (the nvcc compiler) is a separate package, see How to install the CUDA Toolkit on Ubuntu 24.04 / 26.04?.
Most AI software does not need the CUDA Toolkit at all. PyTorch and Ollama only need a working driver.

Frequently asked questions

Which operating system should a GPU server run?
The guides in this category use Ubuntu 24.04 LTS and Ubuntu 26.04 LTS, which NVIDIA lists as supported for its current driver and CUDA releases. Other Linux distributions also work; follow the NVIDIA installation guide for your distribution.

The GPU shows in lspci but I have not installed anything yet. Is that normal?
Yes. A fresh operating system contains no NVIDIA driver until it is installed, so the card is visible but not usable.

How do I see all four GPUs on my multi-GPU server?
Run nvidia-smi or nvidia-smi -L. Each GPU has its own row or line.

Need a GPU server, or a hand with the setup?

GPU dedicated servers: NVIDIA GPU servers for AI training and inference; our engineers can install the driver, CUDA, PyTorch or a private LLM and hand it over ready to use.
Private LLM installation: we install Ollama, the GPU driver and a chat interface on your own server.

Prefer a hand with the setup? Our engineers can do it for you: Hire an Expert, or use our on-demand server management.