[RFC] Deprecate AGP GART support for Radeon/Nouveau/TTM

[Rui Salvaterra]

Hi, Christian,

On Wed, 20 May 2020 at 16:00, Christian König
<ckoenig.leichtzumerken at gmail.com> wrote:
>
> So I've used an ancient system (32bit) to setup a test box for this.
>
>
> The first GPU I could test is an RV280 (Radeon 9200 PRO) which is easily
> 15 years old.

Oh, I have one of those in box somewhere, but no AGP machine to
install it (yet).

> What happens in AGP mode is that glxgears shows artifacts during
> rendering on this system.
>
> In PCI mode those rendering artifacts are gone and glxgears seems to
> draw everything correctly now.
>
> Performance is obviously not comparable, cause in AGP we don't render
> all triangles correctly.

I agree, correctness before performance, always.

> The second GPU I could test is an RV630 PRO (Radeon HD 2600 PRO AGP)
> which is more than 10 years old.
>
> As far as I can tell this one works in both AGP and PCIe mode perfectly
> fine.
>
> Since this is only a 32bit system I couldn't really test any OpenGL game
> that well.

I usually test with my distro's (Debian or Ubuntu, in my case) games.
For example, I used Nexuiz when wiring up the shader cache on r300g.

> But for glxgears switching from AGP to PCIe mode seems to result in a
> roughly 5% performance drop.
>
> The surprising reason for this is not the better TLB performance, but
> the lack of USWC support for the PCIe GART in radeon.
>
> So if anybody wants to get his hands dirty and squeeze a bit more
> performance out of the old hardware, porting USWC from amdgpu to radeon
> shouldn't be to much of a problem.

Well, FWIW, I would argue that a regression in performance, if
avoidable, should be avoided. I have not nearly enough knowledge of
the driver to do it myself, but I'll glady test any patches, on both
x86-64 (Radeon Xpress 1250) and ppc32 (Mobility Radeon 9550).

Cheers,
Rui