mutrace and tiles3

Boudewijn Rempt boud at valdyas.org
Wed Sep 16 11:28:00 CEST 2009


On Wed, 16 Sep 2009, Thomas Zander wrote:

> On Wednesday 16. September 2009 10.51.38 Cyrille Berger wrote:
> > On Wednesday 16 September 2009, Boudewijn Rempt wrote:
> > > I did exactly the same thing as with tiles1. It was much slower -- and
> > > the results are certainly puzzling, since the main contention doesn't
> > > seem to come from the tile engine!
> >
> > I wouldn't be so sure, since 190168 is a mutex used by the event loop, if
> > tiles3 use more signals it can be the explanation.
> 
> I've seen some applications crawl because they generate too many events. The 
> Qt event queue then ends up being an overhead thats too great.  Things like 
> timers are guilty of that and, like Cyrille said, signals among various 
> other things.
> It might be fun to check the amount of events that are executed on the event 
> loop per second coming from Krita vs from X11.
> See QCoreApplication::setEventFilter()

Hm, yes, that would be interesting. Especially since we use signals/slots between
threads for instance for recompositing and progress reporting, and those are events,
too. Only -- that is not a difference between the two tiles engines, and since that's
the only difference in my test, I'm still puzzled.

But I'll investigate that lead, too, it's an excellent idea.

Boudewijn



More information about the kimageshop mailing list