Exa and radeon

[Eric Anholt]

On Thu, 2006-04-27 at 14:58 +0200, Francesco Biscani wrote:
> On Thursday 20 April 2006 18:27, you wrote:
> > On Tue, 2006-04-18 at 15:46 +0200, Francesco Biscani wrote:
> > > On Monday 17 April 2006 22:36, Dario Laera wrote:
> > > > Hi all,
> > > >
> > > > after I've installed xorg-server-7.1_RC1 I've tried to run xcompmgr -a
> > > > with exa to see if some things were improved since 7.0, but I still get
> > > > strange result:
> > > >
> > > > http://laera.web.cs.unibo.it/exa.png
> > > >
> > > > I have a mobility 9600 running on a powerbook.
> > > >
> > > > Thanks, Dario.
> > >
> > > (I'm CC'ing Eric Anholt, since he's the man for exa)
> > >
> > > Hi,
> > >
> > > from what I gather following CVS logs you should add an extra option to
> > > xorg.conf to enjoy the improved exa support in latest xorg:
> > >
> > > Option      "MigrationHeuristic" "always"
> >
> > This is definitely not recommended.  It's just for debugging.  That's
> > why it's undocumented.
> >
> > > So, (that's a question for Eric ;) am I doing something wrong? Can I help
> > > in debugging this?
> >
> > This is probably just driver brokenness.  We're exercising the driver in
> > more ways than before, and I think radeon has issues.
> 
> Hi Eric,
> 
> sorry to bother you again; I just wanted to report that with latest CVS (the 
> one that has "smart" as default migration heuristic) I get the same simptoms 
> as with the "always" migration heuristic (which I talked to you about in a 
> previous mail). I.e., corruption in window titles and slowness in drawing the 
> desktop background in KDE. However, once the background is rendered, moving 
> windows around is _really_ snappy and transparencies work very well :)
> 
> As always, if needed I'm willing to test patches and provide feedback.

There are several rendering issues with Radeon EXA support.  I try to do
EXA infrastructure against fakexa first so that I can trust the
correctness, and then do performance testing on the Radeon.  Even with
current CVS as of a few minutes ago, there's bug #6755 and bug #6756
outstanding for radeon.

I just fixed some migration issues, so hopefully performance will be
better now.  I'm still crawling through fallback logs trying to track
down further performance issues.  I also really need to spend more time
on KDE -- my usual debug session is metacity, mozilla, gnome-terminals,
and xcompmgr -c.

The next patch I'm looking at for EXA is:
https://bugs.freedesktop.org/show_bug.cgi?id=6772

-- 
Eric Anholt                     anholt at FreeBSD.org
eric at anholt.net                 eric.anholt at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 187 bytes
Desc: This is a digitally signed message part
URL: <http://lists.x.org/archives/xorg/attachments/20060427/a7d11106/attachment.pgp>