[cairo] update on cairo-gl performance
Eric Anholt
eric at anholt.net
Tue Oct 20 15:56:34 PDT 2009
Just thought I'd throw out some new numbers here in case people were
curious. This is 2.6.32-rc4, cairo master, libdrm master, mesa master,
2d master, pixman from august 12th. Hardware is Lenovo T61, GM965, 2G
ram, 2Ghz Core 2 Duo.
Note that this isn't apples-to-apples yet. cairo-gl doesn't do
component alpha, so any subpixel text in these tests won't be rendered
correctly. I'm not sure how the setup is supposed to work in a cairo
backend, so I haven't fixed it yet. Anyone understand that?
I've done most of my profiling on firefox-talos-gfx, and through fixes
mostly in Mesa and libdrm I've got it down from upwards of 180 seconds
on this hardware to 56 seconds. At this point, the test consumes around
90% of the CPU still, while around 10% of the time it's idle blocked on
the hardware due to the useless clears of temporary buffers.
What's coming up in cairo-gl:
* cworth is working on GLSL. This will reduce the CPU usage for
state updates, and give us the framework for supporting
gradients and possibly even better shaders for other compositing
than we're doing today with the old nasty fixed function
pipeline.
* ickle is going to be avoiding the extra clears of buffers that
are used internally in cairo. This should be a good win for
both xlib and gl.
* I'm going to keep working on performance outside of cairo. My
next stop will be the gvim trace, I think. Something's going
terribly wrong with memory consumption in this one.
* Once I've got some piglit testcases (should be soon), I'll add
support for GL_ALPHA fbos to the intel GL drivers, which should
cut memory bandwidth consumption for our masks.
Finally, the numbers:
[ # ] backend test min(s) median(s) stddev. count
[ 0] image firefox-talos-gfx 44.507 44.591 0.13% 3/3
[ 0] gl firefox-talos-gfx 56.005 56.231 0.37% 3/3
[ 0] xlib firefox-talos-gfx 92.570 92.631 0.15% 3/3
[ 1] image gvim 7.421 7.438 0.19% 3/3
[ 1] gl gvim 91.088 91.207 0.38% 3/3
[ 1] xlib gvim 49.632 49.638 0.29% 3/3
[ 2] image firefox-planet-gnome 37.715 37.952 0.53% 3/3
[ 2] gl firefox-planet-gnome 91.796 92.135 0.29% 3/3
[ 2] xlib firefox-planet-gnome 48.483 48.596 0.19% 3/3
[ 3] image evolution 6.148 6.156 0.28% 3/3
[ 3] gl evolution 66.187 66.418 1.43% 3/3
[ 3] xlib evolution 41.207 41.226 0.98% 3/3
[ 4] image poppler 5.968 5.978 1.09% 3/3
[ 4] gl poppler 16.815 16.843 3.98% 3/3
[ 4] xlib poppler 35.851 35.888 0.16% 3/3
[ 5] image firefox-talos-svg 74.505 76.889 1.48% 3/3
[ 5] gl firefox-talos-svg 111.540 111.883 1.00% 3/3
[ 5] xlib firefox-talos-svg 96.115 96.513 0.24% 3/3
[ 6] image gnome-system-monitor 8.001 8.057 0.35% 3/3
[ 6] gl gnome-system-monitor 7.865 7.891 0.55% 3/3
[ 6] xlib gnome-system-monitor 9.322 9.547 2.79% 3/3
[ 7] image gnome-terminal-vim 8.504 8.520 0.20% 3/3
[ 7] gl gnome-terminal-vim 14.916 14.956 1.03% 3/3
[ 7] xlib gnome-terminal-vim 44.785 44.864 0.95% 3/3
[ 8] image swfdec-giant-steps 3.039 3.047 0.21% 3/3
[ 8] gl swfdec-giant-steps 10.318 10.391 0.34% 3/3
[ 8] xlib swfdec-giant-steps 6.883 6.929 0.57% 3/3
[ 9] image swfdec-youtube 7.953 8.015 0.42% 3/3
[ 9] gl swfdec-youtube 9.134 9.215 1.52% 3/3
[ 9] xlib swfdec-youtube 5.417 5.473 0.52% 3/3
--
Eric Anholt
eric at anholt.net eric.anholt at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
Url : http://lists.cairographics.org/archives/cairo/attachments/20091020/63c5d2af/attachment.pgp
More information about the cairo
mailing list