NVIDIA just announced something quite interesting, in the form of an upcoming update for their GPU drivers for Linux.
In a post on the official NVIDIA forum titled "DRM Per-Plane Color Pipeline API Support for NVIDIA Open GPU Kernel Modules", they noted how the Linux kernel now has a Colour Pipeline API available as of November 2025. NVIDIA said they were involved with this early on to ensure it would work with their hardware.
From their post:
The goal of the color pipeline API is to allow compositors to leverage GPU display hardware capabilities to accelerate color processing (most notably for HDR) typically done in software. Described as a “prescriptive API,” the API exposes an abstraction of the hardware blocks directly to clients (i.e. Wayland compositors), such that they can directly configure the hardware as they see fit, matching it with their software processing and allowing software/hardware processing to be toggled at will with minimal visible change. Given this API design and the unique constraints of NVIDIA’s display hardware, we expect that compositors will require additional implementation to make effective use of NVIDIA color pipelines.
To help facilitate compositor developers, we are providing a preview of our planned support (official support pending a future driver release) backported against open-gpu-kernel-modules 595.58.03. The NVIDIA color pipeline structure and associated limitations are documented in the commit message at https://github.com/AlexGoinsNV/open-gpu-kernel-modules-drm-color-pipeline-preview/commit/c6d0aefd0d4c. Applying the change on top of the open-gpu-kernel-modules repository will break compatibility with 595.58.03’s X driver, so it should only be used when testing with Wayland + DRM/KMS. To try it out, clone https://github.com/AlexGoinsNV/open-gpu-kernel-modules-drm-color-pipeline-preview and follow the instructions for building/installing against 595.58.03.
kwin-wayland has a known issue that is exposed when running with an NVIDIA driver built with DRM per-plane color pipeline API support. We have documented the issue and recommended a fix here: https://invent.kde.org/plasma/kwin/-/merge_requests/9042.
One interesting note here at the bottom, is that NVIDIA have been using AI tools to help with this with nearly all the code coming from generative AI. Something that perhaps a number of readers won't be too happy to see. In the post they mentioned:
During development, we used Claude Sonnet/Opus to dramatically reduce the time required to create production quality code. Nearly all of the code was produced by the model, but with a strong emphasis on explicit human direction, review, and iteration.
Source: NVIDIA




How to setup OpenMW for modern Morrowind on Linux / SteamOS and Steam Deck
How to install Hollow Knight: Silksong mods on Linux, SteamOS and Steam Deck