intel 2.0 driver >2048x2048 without DRI

[Roland Scheidegger]

m h wrote:
> On 7/19/07, Chase Douglas <chasedouglas at gmail.com> wrote:
>> I've resigned myself to the fact that I am not going to see 3D
>> acceleration with my setup, but I have a laptop display at 1400x1050 and
>> a second monitor at 1600x1200. Is it possible to use the new driver with
>> xrandr v1.2 at all with these screen resolutions?
>>
> 
> I'm using it on my R52 (1400x1050) laptop.  I can do dual screen with
> by 24 inch (1920x1200) montior as well as cloning.  xrandr makes
> controlling the output from the command line quite easy.
> 
> Sadly DRI does NOT work...  I wish this were addressed.
Hmm, it just occured to me that this should be quite easy to make to 
work with private backbuffers with some limitations, in fact I think 
this should basically already work. You could try that out (mesa 
i915tex_privbuffers branch), though you'd need to change the ddx to 
allow dri (and make sure you don't use pageflip).
Limitation would basically be that all of your 3d windows must not 
exceed 2048 pixel in any direction (shouldn't matter where on the screen 
they are - good enough for quake3, probably of no use to compositing 
managers... The dri driver will NOT handle larger windows gracefully, it 
will fall over some way or another (an easy solution without handling 
that for real would be to just never resize the private buffers to 
larger sizes internally, but that's probably not what you'd want). And 
you will need a setup (ddx/drm) which supports the i915tex dri driver.
Disclaimer: I've no idea if that could actually work, or if there are 
other limits coming into play preventing that. The i915tex_privbuffers 
branch was not intended for that...

Roland