Mercurial > libavcodec.hg
graph
-
switch back to (amvd>2)+(amvd>32), its 5 cpu cycles faster now. libavcodecWed, 24 Feb 2010 18:16:48 +0000, by michael
-
Factorize common code from the top of decode_cabac_mb_mvd() libavcodecWed, 24 Feb 2010 18:06:02 +0000, by michael
-
Replace mvd>2 + mvd>32 by MIN((mvd+28)*17>>9, 2) libavcodecWed, 24 Feb 2010 16:16:08 +0000, by michael