News: 0001548426

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Microsoft Lands 62k Lines Of Code Patch In Mesa: Adds New "MFT" Gallium3D Frontend

([Microsoft] 4 Hours Ago Microsoft MFT Gallium3D)


Microsoft's open-source code contributions to the Mesa 3D graphics stack continues... Hitting Mesa 25.2-devel today was a patch adding 61,925 lines of code patch as they introduce a new Gallium3D front-end.

Microsoft contributes to the Mesa graphics driver stack where it benefits their interests for getting more APIs implemented atop Direct3D 12 hardware drivers as well as where it ties into their Windows Subsystem for Linux (WSL) support. Their latest upstream Mesa contribution is introducing the MFT Gallium3D front-end for tieing into their Windows use-cases.

Merged for Mesa 25.2-devel is the Media Foundation Transforms "MFT" front-end and tieing it into their Direct3D 12 (D3D12) Gallium3D driver.

The [1]Microsoft merge request explains:

"This MR contains the implementation of an async media foundation transform (MFT) that uses the D3D12 gallium driver through the pipe interface.

The new mediafoundation frontend code contains the MFT code in a single commit in the PR (more details can be found in the readme.md file in this commit)."

Microsoft's [2]documentation describes Media Foundation Transforms as:

"Media Foundation transforms (MFTs) provide a generic model for processing media data. MFTs are used for decoders, encoders, and digital signal processors (DSPs). In short, anything that sits in the media pipeline between the media source and the media sink is an MFT.

For most applications, the details of MFT data processing are hidden by higher layers of the Media Foundation architecture. Many Media Foundation applications will never make a direct call to an MFT. However, it is certainly possible to host an MFT directly in your application.

MFTs are an evolution of the transform model first introduced with DirectX Media Objects (DMOs). In fact, it is relatively easy to create a transform that supports both models. Compared with DMOs, the required behaviors of MFTs are more clearly specified, which makes it easier to write a correct implementation. In addition, MFTs can support hardware-accelerated video processing."

As of an hour ago, the 62k lines of code patch is [3]merged and now present for enhancing the Mesa stack on Windows.



[1] https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34843

[2] https://learn.microsoft.com/en-us/windows/win32/medfound/media-foundation-transforms

[3] https://cgit.freedesktop.org/mesa/mesa/commit/?id=d348fd5fb5b50edfbdf00f12fd900f77fd658a52



Danny3

RustSystemdWayland

c117152

Volta

bearoso

_r00t-

the-burrito-triangle

NateHubbard

The human mind treats a new idea the way the body treats a strange
protein -- it rejects it.
-- P. Medawar