How to Check for NVIDIA GPU and Set It Up in Ubuntu

If you want to run deep learning models or GPU-accelerated tasks on Ubuntu, having an NVIDIA GPU and the correct drivers is essential. This guide will help you check if a GPU is available and set it up properly for PyTorch.

1 Check if your system has an NVIDIA GPU

Open a terminal and run:

lspci | grep -i nvidia
✅ If output appears, your system has an NVIDIA GPU. 
0000:01:00.0 3D controller: NVIDIA Corporation GP107M [GeForce MX350] (rev a1)
❌ If no output, your system either doesn't have a GPU or it's disabled in BIOS.
💡 Tip: Take note of your GPU model. You'll need it to choose the correct driver.

2 Check if NVIDIA drivers are installed

Run:

nvidia-smi
✅ If working, you will see a table with GPU info, driver version, and CUDA version: 
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 580.95.05   Driver Version: 580.95.05   CUDA Version: 13.0       |
| GPU  Name                 Memory-Usage | GPU-Util  Compute M.               |
| 0   NVIDIA GeForce MX350   5MiB / 2048MiB | 0%                           |
+-----------------------------------------------------------------------------+
❌ If it fails, the driver is not installed or not loaded.

3 Find the recommended NVIDIA driver

Run:

ubuntu-drivers devices

Example output:

driver   : nvidia-driver-580 - distro non-free recommended
⚠️ Important: Always install the recommended driver for your GPU.

4 Install the NVIDIA driver

sudo apt update
sudo apt install nvidia-driver-580

Then reboot your system:

sudo reboot

5 Verify the driver and CUDA

After reboot, check again:

nvidia-smi

You should now see your GPU detected along with driver and CUDA version.

6 Install PyTorch with GPU support

Now that your GPU is working, install PyTorch. Make sure to match the CUDA version from nvidia-smi.

Example for CUDA 13.0:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu130

7 Verify PyTorch GPU access

Open Python:

import torch

print("CUDA Version:", torch.version.cuda)
print("GPU Available:", torch.cuda.is_available())
print("GPU Name:", torch.cuda.get_device_name(0))

Expected output:

CUDA Version: 13.0
GPU Available: True
GPU Name: NVIDIA GeForce MX350
💡 Tip: If torch.cuda.is_available() returns False, check that the driver is loaded properly and your CUDA version matches.

8 CPU-only fallback

If you don't have a GPU or cannot install drivers:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

You can still run PyTorch on CPU, but it will be slower.

Summary

  1. Check for GPU: lspci | grep -i nvidia
  2. Check driver: nvidia-smi
  3. Install recommended driver: ubuntu-drivers devices → sudo apt install <driver>
  4. Verify GPU access
  5. Install PyTorch for GPU (match CUDA version) or CPU-only

With this setup, your Ubuntu machine is ready for GPU-accelerated deep learning.

Comments

Post a Comment

What is your thought about this?

Popular posts from this blog

Use the Google Custom Search JSON API to retrieve images

 To use the Google Custom Search JSON API to retrieve images, you need two things : ✅ 1. Google API Key 🔧 Steps: Go to the Google Cloud Console. Create a new project (or select an existing one). In the left sidebar, go to APIs & Services > Credentials . Click "Create credentials" > "API key" . Copy the API Key shown — you'll use this in your code. ✅ 2. Custom Search Engine ID (cx) 🔧 Steps: Go to the  Custom Search Engine (CSE) page. Click "Get Started" . Enter a domain to search (e.g., www.google.com or leave it as *.com to search the whole web). Click "Create" . After it's created, go to the Control Panel . Under "Search engine ID" , you'll see your cx value — copy this . 🛠 Enable Image Search: In the Custom Search Engine settings (Control Panel): Go to "Basics" > Sites to search → make sure it’s set to “Search the entire web” . Go to ...
Read more

Terminal commands relared to files in the Ubuntu

 🔍 Viewing and Navigating Files Command Description ls List files and directories in the current directory ls -l Long listing (with permissions, size, date, etc.) ls -a Show hidden files (those starting with . ) cd [directory] Change directory pwd Print current working directory tree View directory structure in tree form (may need sudo apt install tree ) 📁 Creating and Managing Files Command Description touch filename Create a new empty file nano filename Edit a file using nano text editor cat filename View file contents more filename / less filename Scroll through file contents cp file1 file2 Copy a file mv oldname newname Rename or move a file rm filename Delete a file rm -i filename Delete with confirmation rm -f filename Force delete without confirmation 📂 Working with Directories Command Description mkdir directory_name Create a directory mkdir -p path/to/directory Create nested directories rmdir directory_name Remove empty directory rm -r directory_name Remove direc...
Read more

Laravel Best Practices: Building Robust and Maintainable Applications

Image
Laravel is one of the most popular PHP frameworks due to its elegance, simplicity, and developer-friendly syntax. But to fully harness Laravel’s power, developers must adhere to certain best practices that ensure their applications are secure, maintainable, and scalable. In this article, we’ll explore typical best practices for Laravel , drawing examples from a real-world scenario—an admin system handling borrow requests. 1. Use Service Classes for Business Logic Why? Keeping your controllers thin and offloading business logic to service classes promotes separation of concerns and cleaner code. public function approveRequestAction($borrowId): JsonResponse { $this->borrowService->approve($borrowId); return response()->json(['message' => 'Request approved']); } 2. Exception Handling Graceful error handling provides better developer and user experience. Use try-catch blocks and Laravel’s built-in exceptions like ModelNotFoundException . try { $th...
Read more