[cairo] [PATCH] Performance patch for bo intersections

Baz brian.ewins at gmail.com
Tue Dec 12 15:39:46 PST 2006


On 12/12/06, Behdad Esfahbod <behdad at behdad.org> wrote:
> Agreed.  I was wondering how much does this specific part costs, given
> that it's more complex than the other tests.  So having them as separate
> commits helps.

Ok, I'll do this. You can try it for yourself after I separate them,
but it seemed to me like the bounding box was by far the bigger win.
This surprised me somewhat as I expected the world map to hit the
coincident point check a lot - and it does - but with hindsight, it
wasn't going to make as much of a difference because rendering and
qsort dominate the profile for that test.

> I also (Oops, should I start a new mail? ;)) want to suggest that these
> be added into a separate function to keep the heuristics separate from
> the real thing.

I'll try that, and all of Carl's suggestions too, thanks for the encouragement.

-Baz


More information about the cairo mailing list