Add Xfermode bench. Also clear before rendering in bench (rather than after).

Add Xfermode bench. Also clear before rendering in bench (rather than after).

Committed: http://code.google.com/p/skia/source/detail?r=8454

[bsalomon, 2013-03-30 01:51:17]

On my home iMac:
whiteoak:skia1 bsalomon$ out/Release/bench --match Xfer --repeat 50
skia bench: alpha=0xFF antialias=1 filter=0 deferred=no logperiter=0 rotate=0
scale=0 clip=0 min=0 record=0 picturerecord=0 dither=default strokeWidth=none
scalar=float system=MAC
njd bench
running bench [640 480]              Xfermode_Darken  8888: cmsecs = 471.50  
565: cmsecs = 584.28   GPU: cmsecs =  41.89 gmsecs =  40.98  NULLGPU: cmsecs =  
3.35
running bench [640 480]             Xfermode_SrcOver  8888: cmsecs =  18.81  
565: cmsecs =  67.52   GPU: cmsecs =   1.24 gmsecs =   5.10  NULLGPU: cmsecs =  
0.74
running bench [640 480]                 Xfermode_Src  8888: cmsecs =   9.28  
565: cmsecs = 212.00   GPU: cmsecs =   1.31 gmsecs =   5.58  NULLGPU: cmsecs =  
0.76

I ran with 10 rects to get a decent amount of overdraw, dumped the images, and
inspected them. 8888 and GPU look the same for all modes. Src mode is slow in
565 and also renders incorrectly.

[robertphillips, 2013-03-31 16:44:13]

lgtm + 1 question/suggestion

https://codereview.chromium.org/13334011/diff/5001/bench/XfermodeBench.cpp
File bench/XfermodeBench.cpp (right):

https://codereview.chromium.org/13334011/diff/5001/bench/XfermodeBench.cpp#ne...
bench/XfermodeBench.cpp:61: 
What about just adding a bench for each xfermode (i.e., why stop at 3)?

[commit-bot: I haz the power, 2013-04-01 12:51:30]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/bsalomon@google.com/13334011/5001

[commit-bot: I haz the power, 2013-04-01 12:51:40]

Change committed as 8454