Daniel Vetter, an engineer at Intel and one of the contributors to the DRM subsystem, has published a roadmap for advancing patches for the Xe driver, designed for use with GPUs based on Intel's Xe architecture, which is utilized in the Arc family graphics cards and integrated graphics starting with Tiger Lake processors. The Xe driver is positioned as a foundation for supporting new chips, independent of code for older platforms. Throughout 2023, the patches are expected to be refined for testing by enthusiasts, as well as potentially reviewed by developers not affiliated with Intel. In an optimistic scenario, the driver may be accepted into the mainline kernel by the end of the year.
Currently, inclusion in the mainline kernel is hindered by the fact that the code was originally developed behind closed doors in a separate kernel branch and now requires additional work to integrate it with the current codebase. Given the recent negative experience with the prolonged integration of significant changes for the AMD driver, which led to the need for rewriting parts of the code, a preliminary consensus on the implementation of the scheduler and interaction with other drivers is proposed to facilitate the advancement of the prepared Xe driver codebase into the main kernel.
The Xe driver is built using a new architecture that makes greater use of existing components within the DRM (Direct Rendering Manager) subsystem, as well as standard components from the i915 driver that are not tied to specific GPUs, such as code for interacting with displays, memory models, and execbuf implementation. In the Xe and i915 drivers, there are plans to ensure the sharing of common code in order to avoid duplicating standard components. In Mesa, OpenGL and Vulkan functionality over the Xe driver is implemented at the level of changes made to the existing Mesa drivers Iris and ANV.
Source: opennet.ru
