Xfermode: SSE2 implementation of softlight_modeproc

Xfermode: SSE2 implementation of softlight_modeproc

With SSE2 optimization, performance of Xfermode_SoftLight will improve
about 30% on desktop i7-3770. Here are the data:
before:
Xfermode_SoftLight   8888:  cmsecs = 379.44   565:  cmsecs =  387.74
after:
Xfermode_SoftLight   8888:  cmsecs = 272.29   565:  cmsecs =  284.31

BUG=skia:

Committed: http://code.google.com/p/skia/source/detail?r=14376

[qiankun, 2014-04-15 07:32:37]

PTAL.

[mtklein, 2014-04-24 17:24:11]

lgtm

https://codereview.chromium.org/236363012/diff/1/src/opts/SkXfermode_opts_SSE...
File src/opts/SkXfermode_opts_SSE2.cpp (right):

https://codereview.chromium.org/236363012/diff/1/src/opts/SkXfermode_opts_SSE...
src/opts/SkXfermode_opts_SSE2.cpp:96: static inline __m128i
SkSqrtBits_SSE2(const __m128i& x, int count) {
Think it's worth spinning off an SkMath_SSE2?

[qiankun, 2014-04-25 08:10:22]

The CQ bit was checked by qiankun.miao@intel.com

[commit-bot: I haz the power, 2014-04-25 08:10:46]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/qiankun.miao@intel.com/236363012/20001

[qiankun, 2014-04-25 08:13:05]

https://codereview.chromium.org/236363012/diff/1/src/opts/SkXfermode_opts_SSE...
File src/opts/SkXfermode_opts_SSE2.cpp (right):

https://codereview.chromium.org/236363012/diff/1/src/opts/SkXfermode_opts_SSE...
src/opts/SkXfermode_opts_SSE2.cpp:96: static inline __m128i
SkSqrtBits_SSE2(const __m128i& x, int count) {
On 2014/04/24 17:24:11, mtklein wrote:
> Think it's worth spinning off an SkMath_SSE2?

Done. Create a new file SkMath_opts_SSE2.h to hold math related functions.

[commit-bot: I haz the power, 2014-04-25 08:34:48]

Change committed as 14376