Get rid of mb2b8_xy and b8_stride, change arrays organized based on b8_stride to
authorMichael Niedermayer <michaelni@gmx.at>
Thu, 25 Feb 2010 23:44:42 +0000 (23:44 +0000)
committerMichael Niedermayer <michaelni@gmx.at>
Thu, 25 Feb 2010 23:44:42 +0000 (23:44 +0000)
commit358b5b1a5958fa12b841aefe09e5e81c7d8bd346
treec6210b5aa5107d8e34d04494ac13880e687c6c4c
parentf3a572ca146555e68c8100d27f948c7fcd4a4ce5
Get rid of mb2b8_xy and b8_stride, change arrays organized based on b8_stride to
ones based on mb_stride in h264.
about 20 cpu cycles faster overall per MB

Originally committed as revision 22065 to svn://svn.ffmpeg.org/ffmpeg/trunk
libavcodec/error_resilience.c
libavcodec/h264.c
libavcodec/h264.h
libavcodec/h264_direct.c
libavcodec/h264_mvpred.h
libavcodec/mpegvideo.c