NVIDIA Finally Embraces Open Source with GPU Kernel Modules

By

With an announcement that sent shockwaves through the open source community, NVIDIA announced it is finally open sourcing its GPU kernel models.

In a recent blog post, NVIDIA announced it is releasing open source GPU kernel modules using a dual GPL/MIT license. This open source release starts with the R515 driver and will continue forward.

According to NVIDIA, this release is "a significant step toward improving the experience of using NVIDIA GPUs in Linux, for tighter integration with the OS, and for developers to debug, integrate, and contribute back."

These new modules will go a long way to improve the out-of-box experience for signing and distributing the NVIDIA driver, so companies like Canonical and SUSE can begin immediately packaging the modules for their distributions. The GPU kernel modules also make it possible to trace code paths and understand how kernel even scheduling is interacting with a workload.

For GPUs in the NVIDIA Turning and Ampere architecture families, the modules are already production-ready, thanks to the rollout of the GSP driver architecture. NVIDIA also plans on adding functionality such as HMM for confidential computing on their NVIDIA Hopper architecture.

Do note, however, that (as of this release), support for GeForce and Workstation GPUs is still in the alpha stages.

Anyone can view the source for the GPU drivers on the NVIDIA GitHub page.

05/16/2022

Related content

comments powered by Disqus
Subscribe to our ADMIN Newsletters
Subscribe to our Linux Newsletters
Find Linux and Open Source Jobs



Support Our Work

ADMIN content is made possible with support from readers like you. Please consider contributing when you've found an article to be beneficial.

Learn More”>
	</a>

<hr>		    
			</div>
		    		</div>

		<div class=