3D MergedFB radeon support for screens > 2048x2048 ?

Alex Deucher alexdeucher at gmail.com
Wed Sep 13 06:42:24 PDT 2006


On 9/13/06, Mitch <Mitch at hasbox.com> wrote:
> Hi,
>
> Screen resolutions of 1280x1024 are pretty much standard these days and
> i note from the radeon man page that 3D is disabled for Mergedfb when
> using two screens with a combined resolution of more than 2048x2048.

the coordinate limits (and tiled surface surface limits) are 2048 on
r100 and r200 based radeons.  It's 4096x4096 theoretically on r300 and
above. Also, 3D isn't disabled, it just does not work on surfaces
wider or tall than 2048 pixels from the start of the visual
framebuffer.  I suppose the man page should be updated to refect the
limits of r100/r200 vis r300 and above.

>
> So what's the solution if you want to use dual screens (bringing the
> resolution upto 2560x2048) with 3D support ? I'm assuming there is a
> technical reason why it's not supported at this res ? Is the limitation
> not there in the propiterary ATI drivers (before i start downloading 56M
> of driver) ? What about using Xinerama ? Any other ideas apart from
> forcing my resolution to 1024x1024 for each screen ?

AFAIK, the ATI driver has the same limitations.  Those are hardware
limits.  If you have an r300 chip or above you should be ok.
Theoretically, if you had a single 3D context that was larger than
2048 in either dimension, you could split it up into tiles and iterate
across them, but it's a lot of work and no one has stepped up to do
it.  Another alternative would be if you didn't need any 3D conects to
be wider/taller than 2048, but you wanted working 3D at offsets
greater than 2048 pixels into the framebuffer would be to update the
3d engine's base address to match the start of the 3D surface rather
than setting it to the framebuffer base as it is now.

I hope that helps,

Alex

>
> Thanks
> M
> _______________________________________________
> xorg mailing list
> xorg at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xorg
>



More information about the xorg mailing list