Improve performance of sse2_combine_over_u()
authorSøren Sandmann Pedersen <ssp@redhat.com>
Fri, 31 Dec 2010 05:57:46 +0000 (00:57 -0500)
committerSøren Sandmann Pedersen <ssp@redhat.com>
Thu, 3 Feb 2011 08:25:05 +0000 (03:25 -0500)
commit8414aa76c20732a6ed29a2d80175936570c5e592
tree57b4283d32935531ca172634a42b42c3f9c330c6
parent08e855f15cba24aac83145b994069d0bb50be5a1
Improve performance of sse2_combine_over_u()

Split this function into two, one that has a mask, and one that
doesn't. This is a fairly substantial speed-up in many cases.

New output of lowlevel-blt-bench over_x888_8_0565:

over_x888_8_0565 =  L1:  63.76  L2:  62.75  M: 59.37 ( 21.55%)  HT: 45.89  VT: 43.55  R: 34.51  RT: 16.80 ( 201Kops/s)
pixman/pixman-sse2.c