Re: [Traverso-devel] playhead cpu usage, crosshair like implementation needed ?

[Remon]

> > These programs make use of opengl, so the driver has to support
> > that!
> >
> > A new acceleration architecture, called EXA is about to become
> > useable, and as far as I understand it accelerates 'normal'
> > painting operations by use of opengl, but I could be mistaken. It
> > should at least be faster then XAA, and EXA should be available in
> > Xorg 7.3 (hopefully lol)
>
> i thought the opengl backend qt provides and their tesselator would
> already do that, or am i mistaken?

Uhm, good question. One can paint on a QGLWidget with a normal painter, but 
I'm not sure how much of that painting will be opengl accelerated?
Hmmm, if it's as much as showed here [1] things look bright when using Qt 
4.3 :-)

Unfortunately, my intel graphicscards needs updated drivers, which are in Xorg 
7.2 which should have been release some monts ago (it hasn't still) so I'm 
unable to test it yet.
But now you remember me of this, a user reported good results using the opengl 
widget, the cpu load reduced quite a bit (~ 40 %), on it's geforece go 7600 
iirc....

Just looked up the exa thing, but it's still a bit unclear to me how it works. 
Oh well, if it works faster, who cares :-)

As for opengl, it's made possible by dri right? The dri memory manager has 
some serious flaws it seems, the good news: There is a new one in Mesa 6.5.2, 
which should boost performance by a good deal so.....

@Nicola: Note that the Audioclip backgrounds are painted transparently, you 
could try to make the color solid again, and see how much it improves things 
on your ppc ?
Does it make sense to add support for solid/transparant audioclip background 
in the theme file?
You could also disable audioclip backgrounds alltogether, and make the 
boundaries a bit more visible :-)

( Something like:
<properties>
        <property name="AudioClip:solidbackground" value="1" />
        <property name="AudioClip:paintbackground" value="0" />
</properties>
)


Remon

1] http://zrusin.blogspot.com/2006/10/benchmarks.html