Installing CUDA and cuDNN on Ubuntu 24.04

This guide walks through the process of installing NVIDIA CUDA Toolkit and cuDNN on Ubuntu 24.04 for deep learning and GPU-accelerated applications.

Prerequisites

Ubuntu 24.04 LTS
A compatible NVIDIA GPU
Sufficient disk space (at least 5GB for CUDA and 1GB for cuDNN)
Administrator (sudo) privileges

Step 1: Verify Your GPU

First, check if your system has a compatible NVIDIA GPU:

lspci | grep -i nvidia

Step 2: Install NVIDIA GPU Drivers

Add the NVIDIA driver repository:

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update

Install the NVIDIA driver (version 570):

sudo apt install libnvidia-common-570 nvidia-driver-570 libnvidia-gl-570

Reboot your system to load the new driver:

sudo reboot

Verify the driver installation after reboot:

nvidia-smi

You should see output showing your GPU and driver version 570.

Step 3: Install CUDA Toolkit 12.8

Download and install the CUDA repository keyring:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update

Install CUDA Toolkit 12.8:

sudo apt-get install cuda-toolkit-12-8

Add CUDA to your PATH by adding these lines to your ~/.bashrc file:

echo 'export PATH=/usr/local/cuda-12.8/bin:$PATH' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda-12.8/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc
source ~/.bashrc

Verify CUDA installation:

nvcc --version

You should see output indicating CUDA version 12.8.

Step 4: Install cuDNN 9.8.0

Follow these installation instructions to install cuDNN:

wget https://developer.download.nvidia.com/compute/cudnn/9.8.0/local_installers/cudnn-local-repo-ubuntu2404-9.8.0_1.0-1_amd64.deb
sudo dpkg -i cudnn-local-repo-ubuntu2404-9.8.0_1.0-1_amd64.deb
sudo cp /var/cudnn-local-repo-ubuntu2404-9.8.0/cudnn-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update

Install cuDNN for CUDA 12:

sudo apt-get -y install cudnn-cuda-12

Reboot your system to finalize the installation:

sudo reboot

Step 5: Verify the Installation

After reboot, verify that both CUDA and cuDNN are installed correctly:

Check CUDA:

nvcc --version

Create a simple test program to verify cuDNN:

cat > cudnn_test.cpp << EOF
#include <cudnn.h>
#include <iostream>

int main() {
    cudnnHandle_t handle;
    cudnnCreate(&handle);
    
    size_t version = cudnnGetVersion();
    std::cout << "cuDNN version: " << version << std::endl;
    
    cudnnDestroy(handle);
    return 0;
}
EOF

nvcc -o cudnn_test cudnn_test.cpp -lcudnn
./cudnn_test

If this runs without errors and displays a cuDNN version, the installation is successful.

Common Issues and Troubleshooting

Missing Libraries

If you encounter missing library errors when running CUDA applications:

sudo apt install build-essential
sudo ldconfig

CUDA and Driver Version Mismatch

Ensure your NVIDIA driver is compatible with CUDA 12.8. If needed, install a specific driver version:

sudo apt install nvidia-driver-xxx

Where xxx is the compatible driver version.

Cannot Find cuDNN

If applications can't find cuDNN libraries, create symbolic links:

sudo ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so.9 /usr/local/cuda-12.8/lib64/libcudnn.so

Verify Environment Variables

Ensure your environment variables are set correctly:

echo $PATH | grep cuda
echo $LD_LIBRARY_PATH | grep cuda

Resources

Note: This guide is specifically for Ubuntu 24.04 with CUDA 12.8 and cuDNN 9.8.0. If you're using different versions, the installation steps might vary.

MarcoDotIO/cuDNN and CUDA Toolkit Ubuntu 24.04 Installation Guide.md