[cairo-bugs] [Bug 10151] Add mediaLib support

[bugzilla-daemon at freedesktop.org]

http://bugs.freedesktop.org/show_bug.cgi?id=10151





------- Comment #18 from daniel.amelang at gmail.com  2007-11-28 15:30 PST -------
(In reply to comment #17)
> I think it shows that many (about 30) functions have 5-14% speedup, and many 
> other functions have smaller speedups. 

Actually, it says 5-14 _times_ faster, not 5-14 percent faster. Don't
shortchange yourself :)

> It seems two functions are slightly slower (paint_similar_rgb_source-512 is
> 6.27x slower and stroke_solid_rgb_source-128 is 4.02x slower though going from
> 9.39 to 9.47 doesn't really seem that significant to me).  There are a few
> other slower functions, all below 1.57x slower, which is probably not
> significant.

Yea, what the heck is going on with cairo-perf calling a 9.39 -> 9.47 change a
"4.02x slowdown"? Carl?

And notice that the top 4 slowdowns are all on the xlib surface, while the
majority of the speedups are on the image surface. The 6x slowdown is pretty
serious, Maybe it has to do with a inefficient xlib fallback-to-image-surface
occuring (especially since the modifications only change pixman, and thus
should only show up when an image surface comes into play). Might this have
something to do with the fact that mlib images are now being created and
destroyed very frequently? So that's something you probably shouldn't ignore.

Along those lines, remember what Carl said here:

https://bugs.freedesktop.org/show_bug.cgi?id=11529#c15

>  Can this now go upstream?

FYI, Soeren is the pixman maintainer, so he makes the final call.

Oh, and you have run the regression tests, right?


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.