Nvidia’s ready to complete their transition to open-source GPU kernel modules on Linux
Nvidia will fully transition to an open-source GPU kernel with upcoming R560 drivers
Nvidia is getting ready to transition to a fully Open Source Linux GPU kernel, starting with their upcoming R560 drivers. Today, Nvidia has achieved equivalent or better GPU performance using their Open Source GPU kernel modules, and contains additional features. As such, Nvidia sees going fully open source with their GPU kernel as a good move.
In 2022, Nvidia started their open source transition with a set of Linux GPU kernel modules. These modules applied to both GeForce and Workstation GPU models on Linux. Soon, Nvidia’s newer GPUs will be supported by the company’s Open Source kernel. This includes Turing, Ampere, Ada Lovelace, Hopper, Grace Hopper, and Blackwell. For cutting-edge Grace Hopper and Blackwell platforms, Nvidia’s new Open Source kernel will be mandatory. For other GPUs, both Open Source and Proprietary kernels will be usable. However, Nvidia recommends switching to the new open source modules.
Unfortunately, Nvidia’s open source kernel modules do not support Nvidia’s older GPUs. Maxwell, Pascal, and Volta based GPUs will need to continue using Nvidia’s proprietary driver.
With the R515 driver, NVIDIA released a set of Linux GPU kernel modules in May 2022 as open source with dual GPL and MIT licensing. The initial release targeted datacenter compute GPUs, with GeForce and Workstation GPUs in an alpha state.
At the time, we announced that more robust and fully-featured GeForce and Workstation Linux support would follow in subsequent releases and the NVIDIA Open Kernel Modules would eventually supplant the closed-source driver.
NVIDIA GPUs share a common driver architecture and capability set. The same driver for your desktop or laptop runs the world’s most advanced AI workloads in the cloud. It’s been incredibly important to us that we get it just right.
Two years on, we’ve achieved equivalent or better application performance with our open-source GPU kernel modules and added substantial new capabilities:
- Heterogeneous memory management (HMM) support
- Confidential computing
- The coherent memory architectures of our Grace platforms
- And more
We’re now at a point where transitioning fully to the open-source GPU kernel modules is the right move, and we’re making that change in the upcoming R560 driver release.
Today, Nvidia’s GPUs are achieving higher performance levels and gaining a broader feature set with their Open Source kernel modules. Now that this has been achieved, it makes little sense for Nvidia to continue supporting their proprietary drivers on Linux. Moving fully to open source just makes sense, especially if it can yield further improvements for Nvidia.
You can join the discussion on Nvidia’s open source GPU kernel plans on the OC3D Forums.
