Xegl lives!

[Jim Gettys]

On Tue, 2005-05-24 at 16:23 -0400, Jon Smirl wrote:

> > 
> > Do you have numbers to back up that claim?  I'd be particularly interested in
> > comparing glitz performance to the normal Render core on Xati or on nVidia's
> > driver.
> 
> The have been dozens of Cairo benchmarks posted to the Cairo group.
> I'm actually being conservative, sometimes glitz wins by 100x.
> Probably best to ask in the cairo group for recent numbers.
> 

Yes, the Glitz backend is the fastest, often by a lot.

And there are other times that Cairo performance still is plain poor, on
all cairo back-ends, that has to be fixed, and the Cairo folks are
finishing up a major API revision before Cairo goes to its 1.0 release
(when API stability becomes guaranteed).  So both the API upheaval and
performance work is where Cairo is spending its time right now.
Hopefully, the dust settles in the next two months or so, according to
the schedules I've seen.

The other reality of 20 years experience has been that you *always* have
to have a software implementation that goes to pixels.  No graphics
system every made has ever done everything all applications want and
they always need a way to get their hands on pixels, for those
unanticipated needs.  Not to mention not all hardware has 3D
capability.....

So much as we'd like to be able to presume that GL is the answer under
all circumstances, it isn't, and won't be for years more into the
future.
				- Jim