Setting Up CUDA for GPU Computing

CUDA (Compute Unified Device Architecture) is a parallel computing platform and application programming interface (API) model created by NVIDIA. It allows developers to use a CUDA-enabled graphics processing unit (GPU) for general purpose processing. This tutorial will guide you through the steps to install and configure CUDA on a Linux system for GPU computing.

Prerequisites

• A Linux-based system (Ubuntu is commonly used).
• CUDA-capable GPU (NVIDIA GPU).
• Root or sudo privileges to install software.

1. Checking GPU Compatibility

First, confirm that your NVIDIA GPU supports CUDA. You can check the list of supported GPUs on the NVIDIA CUDA GPUs page.

2. Installing NVIDIA Drivers

You need to install the appropriate NVIDIA drivers for your GPU. You can install drivers on Ubuntu using the following commands:

sudo apt update
sudo apt install nvidia-driver- -y

Replace <version> with the desired driver version.

After installing the drivers, reboot your system:

sudo reboot

3. Installing CUDA Toolkit

Next, download the CUDA Toolkit from the NVIDIA CUDA Downloads page. Select your operating system, architecture, distribution, and version. Choose the Installer type: deb (network) for Ubuntu.

After downloading, run the following command to install the CUDA Toolkit:

sudo dpkg -i cuda-repo-ubuntu_.deb
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu/x86_64/7fa2af80.pub
sudo apt update
sudo apt install cuda -y

Replace <version> and <deb-package> with the version you downloaded.

4. Setting Environment Variables

After installation, add the CUDA toolkit to your PATH. Open the .bashrc file:

nano ~/.bashrc

Add the following lines at the end of the file:

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

Save the file and apply the changes:

source ~/.bashrc

5. Verifying the Installation

To verify if CUDA has been installed correctly, run the following command:

nvcc --version

This will display the version of nvcc compiled with CUDA. You can also run a sample program to check GPU functionality:

cd /usr/local/cuda/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery

The output will tell you about the details of your GPU if everything is functioning properly.

6. Conclusion

You have successfully set up CUDA for GPU computing on your Linux system. With a properly configured environment, you can now develop applications that take advantage of the powerful parallel computing capabilities offered by NVIDIA GPUs. Explore NVIDIA’s CUDA documentation to leverage its full potential in your projects!

About the Author

Dorian Kane

Author

View All Posts