Installing Cuda

CUDA 12.3 should have already been installed as part of the virtual machine creation.

Check the installation of the NVIDIA CUDA Toolkit

What is the CUDA Toolkit?

The NVIDIA CUDA Toolkit provides a development environment for creating high performance GPU-accelerated applications.

With the CUDA Toolkit, you can develop, optimise, and deploy your applications on GPU-accelerated systems.

The toolkit includes GPU-accelerated libraries, debugging and optimisation tools, a C/C++ compiler, and a runtime library to deploy your application.

If not installed, follow the instructions below.

Installing the NVIDA CUDA Toolkit

First, check to see if the CUDA Toolkit is installed, we can check to see whether the core compiler is installed, the NVIDIA CUDA Compiler (NVCC)

Nvidia CUDA Compiler (NVCC) is a part of the CUDA Toolkit. It is the compiler for CUDA, responsible for translating CUDA code into executable programs.

NVCC takes high-level CUDA code and turns it into a form that can be understood and executed by the GPU.

It handles the partitioning of code into segments that can be run on either the CPU or GPU, and manages the compilation of the GPU parts of the code.

First, to check if NVCC is installed and its version, run

nvcc --version

If the NVIDIA CUDA Toolkit has been installed, the output will look something like this:

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 12.1, V11.8.89
12.1.r11.8/compiler.31833905_0

You should see release 12.3 - which indicates the CUDA Toolkit Version 12.3 has been successfully installed.

Install Cuda 12.3

CUDA Toolkit 12.3 DownloadsNVIDIA Developer

CUDA 12.3

Enter in the requirements and the instructions will appear

Download the CUDA keyring package

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb

• This command uses wget to download the CUDA keyring package from the specified URL.

• The CUDA keyring package is a digital signature package that ensures the authenticity of the CUDA repositories.

This command installs the downloaded CUDA keyring package:

sudo dpkg -i cuda-keyring_1.1-1_all.deb

The dpkg -i flag is used to install the downloaded CUDA keyring package (cuda-keyring_1.0-1_all.deb).

Installing the keyring package adds the CUDA repository's GPG key to the system's keyring, allowing the package manager to verify the integrity of the CUDA packages.

Update package lists from repositories on system including the newly added CUDA repository

sudo apt-get update

This command will then install the CUDA package and its dependencies on your system

sudo apt-get -y install cuda-toolkit-12-3

• The -y flag automatically answers "yes" to any prompts during the installation process, allowing the installation to proceed without user intervention.

• The cuda package is a meta-package that installs the CUDA Toolkit, including the CUDA libraries, runtime, and development tools.

Make sure CUDA is on PATH

Verify the Installation:

Ensure that CUDA was installed correctly without errors. You can check the installation logs for any errors or warnings that might have occurred during the installation process.

Use the command to list all CUDA-related packages installed and their versions. Enter into the terminal:

dpkg -l | grep cuda

This will provide a list of all CUDA related packages:

Below are some example files that this command highlights, for example:

cuda These are meta-packages for CUDA. Installing a meta-package installs all the components of CUDA

cuda-cccl: CUDA CCCL (CUDA C++ Core Library) is part of the CUDA Toolkit and provides essential libraries for CUDA C++ development.

cuda-command-line-tools: This package includes command-line tools for CUDA, such as nvcc (NVIDIA CUDA Compiler), which is crucial for compiling CUDA code.

cuda-compiler: This package includes the CUDA compiler, which is essential for converting CUDA code into code that can run on Nvidia GPUs.

cuda-demo-suite: This package contains demos showcasing the capabilities and features of CUDA.

cuda-documentation: Provides the documentation for CUDA, useful for developers to understand and use CUDA APIs.

cuda-driver-dev: Includes development resources for the CUDA driver, such as headers and stub libraries.

cuda-libraries, cuda-libraries-dev: These meta-packages include libraries necessary for CUDA development and their development counterparts.

cuda-nvcc: NVIDIA CUDA Compiler (NVCC) is a tool for compiling CUDA code.

Check Environment Variables:

• Ensure that your environment variables are pointing to the new CUDA installation. Specifically, check the PATH and LD_LIBRARY_PATH environment variables via the following commands:

echo $PATH

echo $LD_LIBRARY_PATH

Update these variables if they are pointing to an older CUDA version.

If CUDA is not on PATH:

Export PATH

export PATH=/usr/local/cuda-12.3/bin${PATH:+:${PATH}}

This command adds /usr/local/cuda-12.3/bin to the PATH environment variable.

PATH is a list of directories the shell searches for executable files. Adding CUDA's bin directory makes it possible to run CUDA tools and compilers directly from the command line without specifying their full path.

${PATH:+:${PATH}} is a shell parameter expansion pattern that appends the existing PATH variable to the new path. If PATH is unset, nothing is appended.

LD_LIBRARY_PATH Variable Setup

Export LD_LIBRARY_PATH for 64-bit OS:

export LD_LIBRARY_PATH=/usr/local/cuda-12.3/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

LD_LIBRARY_PATH is an environment variable specifying directories where libraries are searched for first, before the standard set of directories.

This command adds the lib64 directory of the CUDA installation to LD_LIBRARY_PATH, which is necessary for 64-bit systems. It ensures that the system can find and use the CUDA libraries.

Verify NVCC Version:

After updating the environment variables, check the NVCC version again using:

nvcc --version

This should reflect the new version if the environment variables are set correctly.

Update Alternatives:

• Sometimes, multiple versions of CUDA can coexist, and the system may still use the old version. Use update-alternatives to configure the default CUDA version.

• Run sudo update-alternatives --config nvcc to choose the correct version of NVCC.

sudo update-alternatives --config

After the installation is complete, you will have the CUDA Toolkit and runtime installed on your Ubuntu environment, enabling you to develop and run CUDA-accelerated applications using NVIDIA GPUs.