Nvidia has announced that it will open Linux drivers for its graphics cards, starting from version R515, using a dual GPL/MIT license. Source code for kernel modules will be available in the NVIDIA Open GPU Kernel Modules repository on GitHub, but at this time, only data center GPU code is considered production-ready. GeForce and Workstation GPUs are currently considered “alpha quality”.
With fully built packages available, along with source code, the move will make it easier for distribution managers to include the drivers in their software repositories, with both Canonical (makers of Ubuntu) and SUSE named as developers who can now package open kernel modules with their distributions. Nvidia worked with both companies, as well as Red Hat, to improve package deployment and create better support models for customers.
Nvidia says code is currently production-ready for Turing and Ampere family data center GPUs following the gradual rollout of the GSP driver architecture over the past year. It has been tested on a wide variety of workloads to ensure feature and performance parity with the proprietary kernel-mode driver, but also brings new features, such as the DMA-BUF framework for sharing buffers between peripherals and subsystems that will take on their full meaning. with the Hopper architecture. Anyone running GPUs using an architecture that predates Turing will have to continue using the old-school proprietary drivers.
Clearly, one of Nvidia’s primary goals with these drivers is to improve support and functionality for supercomputers and large data center installations. Every major supercomputer runs on some version of Linux, or so, and having closed-source drivers probably isn’t suitable for the people responsible for those installations.
The picture for home users isn’t quite so rosy at the moment, however, with only the obsolete, proprietary monolithic kernel module being considered anything other than Alpha quality. Improvements are, according to Nvidia, planned throughout the year.
The new open-source kernel-mode driver works with the same firmware as the previous driver and the same user-mode stacks such as CUDA, OpenGL, and Vulkan. The only stipulation is that all components of the driver stack must match versions within a release. Community-submitted fixes will be considered for inclusion in future driver releases.
“New Nvidia open-source GPU core modules will simplify installations and increase security for Ubuntu users, whether they are AI/ML developers, gamers or cloud users,” said Cindy Goldberg, vice president of alliances. Silicon at Canonical. The new drivers should make their way into the recently launched Canonical Ubuntu 22.04 LTS in the coming months.
Nvidia also revealed that it is working on an upstream approach with the Linux kernel community and partners, as its current codebase does not conform to Linux design conventions and is not a candidate for Linux upstream. The source code will also be used to improve the Open Source Nouveau driver.
Developers can download development drivers as part of CUDA Toolkit 11.7. A full data center pilot will follow.