1. Install NVIDIA CUDA Toolkit

The following instruction describes installing CUDA for Ubuntu, CentOS, and Red Hat (RHEL). You may refer to CUDA installation guide from NVIDIA developer documentation for other operating systems and platforms.

1.1. Install CUDA Runtime Libraries#

To download and install the CUDA Toolkit on both Linux and Windows, refer to the NVIDIA Developer website. It’s important to note that NVIDIA’s installation instructions on their website include the entire CUDA Toolkit, which is typically quite large (over 6 GB in size).

However, for running g-learn, you don’t need to install the entire CUDA Toolkit. Instead, only a few of the CUDA runtime libraries, are required. Below are simplified installation instructions for Linux, allowing you to perform a minimal CUDA installation with only the necessary libraries.

1.1.1. Add CUDA Repository#

Before installing CUDA libraries, add CUDA repository to your package manager:

Ubuntu/Debian

# Machine architecture
ARCH=$(uname -m | grep -q -e 'x86_64' && echo 'x86_64' || echo 'sbsa')

# OS Version
UBUNTU_VERSION=$(awk -F= '/^VERSION_ID/{gsub(/"/, "", $2); print $2}' /etc/os-release)
OS_VERSION=$(dpkg --compare-versions "$UBUNTU_VERSION" "ge" "22.04" && echo "2204" || echo "2004")

# Add CUDA Repository
sudo apt update
sudo apt install wget -y
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${OS_VERSION}/${ARCH}/cuda-keyring_1.1-1_all.deb -P /tmp
sudo dpkg -i /tmp/cuda-keyring_1.1-1_all.deb
rm /tmp/cuda-keyring_1.1-1_all.deb

CentOS 7

# Machine architecture
ARCH=$(uname -m | grep -q -e 'x86_64' && echo 'x86_64' || echo 'sbsa')

# OS Version
OS_VERSION=$(awk -F= '/^VERSION_ID/{gsub(/"/, "", $2); print $2}' /etc/os-release)

# Add CUDA Repository
sudo yum install -y yum-utils
sudo yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel${OS_VERSION}/${ARCH}/cuda-rhel${OS_VERSION}.repo

RHEL 9

# Machine architecture
ARCH=$(uname -m | grep -q -e 'x86_64' && echo 'x86_64' || echo 'sbsa')

# OS Version
OS_VERSION=$(awk -F= '/^VERSION_ID/{gsub(/"/, "", $2); print $2}' /etc/os-release)

# Add CUDA Repository
sudo dnf install -y dnf-utils
sudo dnf config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel${OS_VERSION}/${ARCH}/cuda-rhel${OS_VERSION}.repo

1.1.2. Install Minimal CUDA Libraries#

In the following, you may change CUDA_VERSION to the CUDA version that you wish to install.

Ubuntu/Debian

# Set to the desired cuda version
CUDA_VERSION="12-3"

# Install required CUDA libraries
sudo apt-get update
sudo apt install -y \
    cuda-cudart-${CUDA_VERSION} \
    libcublas-${CUDA_VERSION} \
    libcusparse-${CUDA_VERSION}

CentOS 7

# Choose a desired cuda version
CUDA_VERSION="12-3"

# Install required CUDA libraries
sudo yum install --setopt=obsoletes=0 -y \
    cuda-cudart-${CUDA_VERSION} \
    libcublas-${CUDA_VERSION} \
    libcusparse-${CUDA_VERSION}

RHEL 9

# Choose a desired cuda version
CUDA_VERSION="12-3"

# Install required CUDA libraries
sudo dnf install --setopt=obsoletes=0 -y \
    cuda-nvcc-${CUDA_VERSION} \
    libcublas-${CUDA_VERSION} \
    libcusparse-${CUDA_VERSION}

Export LD_LIBRARY_PATH environment variable with the CUDA library location by

echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64${PATH:+:${LD_LIBRARY_PATH}}' >> ~/.bashrc
source ~/.bashrc

1.2. Install NVIDIA Graphic Driver#

First, make sure you have added CUDA repository. Then, install NVIDIA graphic driver with

Ubuntu/Debian

export DEBIAN_FRONTEND=noninteractive
sudo -E apt install cuda-drivers -y

CentOS 7

sudo yum -y install nvidia-driver-latest-dkms

RHEL 9

sudo dnf -y module install nvidia-driver:latest-dkms

The above step might need a reboot afterwards to properly load NVIDIA graphic driver. Confirm the driver installation by

nvidia-smi

1. Install NVIDIA CUDA Toolkit#

1.1. Install CUDA Runtime Libraries#

1.1.1. Add CUDA Repository#

1.1.2. Install Minimal CUDA Libraries#

1.2. Install NVIDIA Graphic Driver#

1.3. Install OpenMP#