Michael Niedermayer 3a9f44d5d5 and of course the unneeded double subtractions were blindly put in the
mmx code
this also makes the affected code 4% faster

Originally committed as revision 10156 to svn://svn.ffmpeg.org/ffmpeg/trunk
2007-08-20 22:29:21 +00:00
..
2007-06-12 09:29:25 +00:00
2007-06-12 18:50:50 +00:00
2007-06-12 18:50:50 +00:00
2007-07-05 12:06:02 +00:00
2007-07-05 12:06:02 +00:00