Xegl lives!

David Reveman davidr at novell.com
Thu May 26 09:01:51 PDT 2005


On Wed, 2005-05-25 at 11:12 -0700, Allen Akin wrote: 
> On Wed, May 25, 2005 at 12:49:49AM -0400, Jon Smirl wrote:
> | Allen, I've always been suspicious that the Cairo imaging model wasn't
> | really compatible with the OpenGL one but I lack the background to
> | make a true comparison. Are the models really compatible?
> 
> I don't know the subtleties of Cairo well enough to give you a good
> answer.  Maybe the Glitz folks will chime in.

Well, there's nothing wrong cairo's imaging model, and it's not like
it's a completely new model, it's very similar to PostScript, PDF, SVG,
Quartz... and I think we're very confident that this is the imaging
model that application developers like to use in the future. There's
nothing that really makes it incompatible with OpenGL. It's just that
the quality requirements are much higher than what OpenGL is usually
used for. With high quality output comes multi-pass rendering algorithms
and use of intermediate offscreen buffers, things OpenGL implementations
never have had good support for. But this will most likely change in the
near future.

The current glitz backend in cairo is not cheating in any way, the
output quality is as good as with the image backend and the performance
gain is no surprise. It's easy to measure the OpenGL blending and image
transformation performance of a card and compare it to a software
renderer. The performance gain you'll see in such a test will be pretty
the same performance gain you'll get from the glitz backend. The high
quality geometry that cairo is producing is the hard part to accelerate
and it's not as fast as we like it to be right now, all due to the
problems I mentioned above, so it will get better soon.

I think it's much better to create an API design for an imaging model
that is hard to accelerate right now but what application developers
want for the future than to create an API that is well accelerated today
but uninteresting to application developers in a year or so.

-David




More information about the xorg mailing list