diff options
author | Michael Niedermayer <michaelni@gmx.at> | 2010-01-17 23:35:55 +0300 |
---|---|---|
committer | Michael Niedermayer <michaelni@gmx.at> | 2010-01-17 23:35:55 +0300 |
commit | c988f97566cdf536ba0dcbc0d77d885456852060 (patch) | |
tree | dcaf443e415311e25f4012d0bc504659e51ed48a /libavcodec/h264_cavlc.c | |
parent | 00c4127ec9ff30caaa4579d2d1ef1557d870a7f1 (diff) |
Rearchitecturing the stiched up goose part 1
Run loop filter per row instead of per MB, this also should make it
much easier to switch to per frame filtering and also doing so in a
seperate thread in the future if some volunteer wants to try.
Overall decoding speedup of 1.7% (single thread on pentium dual / cathedral sample)
This change also allows some optimizations to be tried that would not have
been possible before.
Originally committed as revision 21270 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/h264_cavlc.c')
-rw-r--r-- | libavcodec/h264_cavlc.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/libavcodec/h264_cavlc.c b/libavcodec/h264_cavlc.c index c170e7cc92..700064019c 100644 --- a/libavcodec/h264_cavlc.c +++ b/libavcodec/h264_cavlc.c @@ -620,7 +620,7 @@ decode_intra_mb: // In deblocking, the quantizer is 0 s->current_picture.qscale_table[mb_xy]= 0; // All coeffs are present - memset(h->non_zero_count[mb_xy], 16, 16); + memset(h->non_zero_count[mb_xy], 16, 32); s->current_picture.mb_type[mb_xy]= mb_type; return 0; |