diff options
| author | Jason Garrett-Glaser <jason@x264.com> | 2011-03-11 13:43:10 -0800 |
|---|---|---|
| committer | Jason Garrett-Glaser <jason@x264.com> | 2011-03-12 03:43:42 -0800 |
| commit | 1eeca88691ad0fd232f110f3a389ebb494c0a6dc (patch) | |
| tree | 0171bf1d70aba71084569d79bb54c45607c3ed6a /libavcodec/vp8.c | |
| parent | 3efbe137396d39e225315a9d1e4bf82d1a9c92e5 (diff) | |
| download | ffmpeg-1eeca88691ad0fd232f110f3a389ebb494c0a6dc.tar.gz | |
VP8: optimize VP8Context struct ordering
Shaves at least 3KB off code size on x86, should improve cache utilization.
This would probably be useful to do for other decoders/encoders as well.
Diffstat (limited to 'libavcodec/vp8.c')
| -rw-r--r-- | libavcodec/vp8.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/libavcodec/vp8.c b/libavcodec/vp8.c index de077c8886..42f401d2b1 100644 --- a/libavcodec/vp8.c +++ b/libavcodec/vp8.c @@ -474,7 +474,7 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y) enum { EDGE_TOP, EDGE_LEFT, EDGE_TOPLEFT }; int idx = CNT_ZERO; int cur_sign_bias = s->sign_bias[mb->ref_frame]; - int *sign_bias = s->sign_bias; + int8_t *sign_bias = s->sign_bias; VP56mv near_mv[4]; uint8_t cnt[4] = { 0 }; VP56RangeCoder *c = &s->c; |