QT4 perfomance.

[Eric Anholt]

On Tue, 2005-06-28 at 23:34 +0200, Alexander S. Usov wrote:
> I think I am not the first to notice it, but still I think it's worth 
> asking :)
> 
> Thanks to the final release of qt4 I decided to give it a try and to
> watch some of their samples. I have tried it on 2 machines, one having
> matrox g550 in it, and other with radeon mobility 9000.
> 
> It really sucks :)
> I mean visually it's quite nice, but even in the mainwindow demo (the one
> with a bunch of docked windows in it) it feels really *slow* and eats up
> 100% of cpu. HW acceleration is working on both machines.
> Does anybody knows if it is just due to sloppy coding of the demos, or we
> will have to wait till next xorg release with KAA it to be able to use it?
> The most funny part is that it works like a charm under windows.

If they're using cairo, it's going to be slow.  The cairo folks have
just started on performance, and they tell me there's a lot of
low-hanging fruit.  And XAA makes render acceleration (the operations
that cairo pretty much consists of, in the absence of glitz) really hard
-- factor of 10 performance loss, in the few cases it can accelerate at
all.

Hopefully EXA (the KAA-in-Xorg) will hit the tree soon and we can start
converting drivers -- I started on MGA for practice last night, and it
should be pretty easy.  In the meantime, xorg-server-snap will provide
some limited improvements, thanks to better software fallbacks for
render (from Zack Rusin of trolltech).

-- 
Eric Anholt                                     eta at lclark.edu
http://people.freebsd.org/~anholt/              anholt at FreeBSD.org