diff options
| author | Michael Niedermayer <michaelni@gmx.at> | 2008-12-18 02:53:18 +0000 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2008-12-18 02:53:18 +0000 |
| commit | 2fd1f0e0262da82d05517c8c104dcd9a2007fa0d (patch) | |
| tree | 201833a5a6ea6686f9dddd019b372b47c250d0c1 /libavcodec/h264.c | |
| parent | ac22385931f847cdabde4df8b0f6450d0ad14de1 (diff) | |
| download | ffmpeg-2fd1f0e0262da82d05517c8c104dcd9a2007fa0d.tar.gz | |
Use the new idct functions (except chroma as it was slower in benchmarks)
cathedral +0.5% speed
aladin +0.6% speed [note aladin has been cat-ed 10 times to reduce the influence
of init time]
Speedup also verified via START/STOP_TIMER (difference was very significant
for the changed parts)
Originally committed as revision 16207 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/h264.c')
| -rw-r--r-- | libavcodec/h264.c | 21 |
1 files changed, 14 insertions, 7 deletions
diff --git a/libavcodec/h264.c b/libavcodec/h264.c index e5c9e33611..d15845bc2f 100644 --- a/libavcodec/h264.c +++ b/libavcodec/h264.c @@ -2515,27 +2515,34 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){ if(!IS_INTRA4x4(mb_type)){ if(is_h264){ if(IS_INTRA16x16(mb_type)){ - if(transform_bypass && h->sps.profile_idc==244 && (h->intra16x16_pred_mode==VERT_PRED8x8 || h->intra16x16_pred_mode==HOR_PRED8x8)){ + if(transform_bypass){ + if(h->sps.profile_idc==244 && (h->intra16x16_pred_mode==VERT_PRED8x8 || h->intra16x16_pred_mode==HOR_PRED8x8)){ h->hpc.pred16x16_add[h->intra16x16_pred_mode](dest_y, block_offset, h->mb, linesize); }else{ for(i=0; i<16; i++){ - if(h->non_zero_count_cache[ scan8[i] ]) + if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]) idct_add (dest_y + block_offset[i], h->mb + i*16, linesize); - else if(h->mb[i*16]) - idct_dc_add(dest_y + block_offset[i], h->mb + i*16, linesize); } + } + }else{ + s->dsp.h264_idct_add16intra(dest_y, block_offset, h->mb, linesize, h->non_zero_count_cache); } }else if(h->cbp&15){ + if(transform_bypass){ const int di = IS_8x8DCT(mb_type) ? 4 : 1; for(i=0; i<16; i+=di){ int nnz = h->non_zero_count_cache[ scan8[i] ]; if(nnz){ - if(nnz==1 && h->mb[i*16]) - idct_dc_add(dest_y + block_offset[i], h->mb + i*16, linesize); - else idct_add(dest_y + block_offset[i], h->mb + i*16, linesize); } } + }else{ + if(IS_8x8DCT(mb_type)){ + s->dsp.h264_idct8_add4(dest_y, block_offset, h->mb, linesize, h->non_zero_count_cache); + }else{ + s->dsp.h264_idct_add16(dest_y, block_offset, h->mb, linesize, h->non_zero_count_cache); + } + } } }else{ for(i=0; i<16; i++){ |