Status of ATI video acceleration support?

[Sergey Jamy]

I understand sailing along the path of least resistance...


2013/5/7 Jean-Sébastien Pédron <dumbbell at freebsd.org>

> On 07.05.2013 11:33, Sergey Jamy wrote:
> > Who is this "Radeon X.Org"? I am know X.Org and radeon (AMD).
>
> The complete quote would be "Radeon X.Org driver" and the real name of
> the project is xf86-vidoe-ati:
>     http://www.x.org/wiki/radeon
>     http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/
>
> This corresponds to the userland-side of the driver for AMD GPUs. Before
> the 7.0.0 release, this project comprised two drivers:
>     1. the UMS driver ("User Mode Setting", which we still use on
>        FreeBSD).
>     2. the KMS driver ("Kernel Mode Setting").
>
> Starting from 7.0.0, only the KMS driver remains.
>
> > AMD has official support for freebsd?
>
> No.
>
> > Why we can't have self Radeon X.Org? Whatever we patching their
> > code for compatibility with our kernel and (maybe) stdlibC.
>
> I'm not sure I understand your question. We should not have to patch
> anything in xf86-video-ati (maybe some small changes), because the
> userland part of the driver (libdrm, xf86-video-ati, Mesa) uses standard
> syscalls and DRM-defined ioctls.
>
> > Who prevents us from having your driver? Perhaps even the
> > manufacturer will participate. For example, as was the case with
> > ACPI.
>
> AMD already employs several developers working on the opensource driver
> (both the kernel part in Linux and the userland part in libdrm,
> xf86-video-ati and Mesa). The kernel part is what we're working on for
> FreeBSD. Other OSes are making progress too. The userland part should be
> reusable by non-Linux systems anyway. If it's not the case, we'll work
> with upstream developers for sure.
>
> --
> Jean-Sébastien Pédron
>
>