Canonical announced today that they will formally support the NVIDIA CUDA toolkit and also make it available via the Ubuntu repositories.
As part of their collaboration with NVIDIA, Canonical will be officially supporting and distributing CUDA within Ubuntu. In the age of AI, Canonical is trying to make it super easy to run CUDA on Ubuntu Linux even though the user-space software is closed-source. At least though there is the modern, open-source NVIDIA GPU kernel driver for hardware since Turing.
Canonical announced moments ago on their blog:
“Historically, developers would download the CUDA Toolkit directly from NVIDIA’s website. Today, Canonical is making it even easier for developers to access CUDA natively through their development environment. The CUDA toolkit and runtime will be directly distributed within Ubuntu. Developers using this new distribution channel will be able to use CUDA on their hardware with a native Ubuntu experience. Once CUDA redistribution is fully integrated into Ubuntu, application developers and system administrators can expect the current multi-step CUDA installation process to become a single command.
For application developers targeting Ubuntu systems, this new distribution model means they can simply declare the CUDA runtime, while Ubuntu manages its installation and compatibility across a wide range of supported NVIDIA hardware. This ensures that CUDA will be more accessible and integrated into a widely-used and trusted Linux distribution.”
AMD for their part is also working to make it easier to deploy the ROCm compute stack on Ubuntu (and other Linux distributions) as well.
