diff options
| author | Jason Garrett-Glaser <darkshikari@gmail.com> | 2011-01-14 21:34:25 +0000 |
|---|---|---|
| committer | Jason Garrett-Glaser <darkshikari@gmail.com> | 2011-01-14 21:34:25 +0000 |
| commit | 19fb234e4af1ff9f58ff2fdd604ac6f6bb87ad6b (patch) | |
| tree | 220be84d79d9c771c1afeab43fdd2aaa82fea01d /libavcodec/h264_cabac.c | |
| parent | 6c18f1cda2e2b2471ebf75d30d552cb0cb61b6ad (diff) | |
H.264: split luma dc idct out and implement MMX/SSE2 versions
About 2.5x the speed.
NOTE: the way that the asm code handles large qmuls is a bit suboptimal.
If x264-style dequant was used (separate shift and qmul values), it might
be possible to get some extra speed.
Originally committed as revision 26336 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/h264_cabac.c')
| -rw-r--r-- | libavcodec/h264_cabac.c | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/libavcodec/h264_cabac.c b/libavcodec/h264_cabac.c index 485837879d..971af37114 100644 --- a/libavcodec/h264_cabac.c +++ b/libavcodec/h264_cabac.c @@ -1597,17 +1597,15 @@ decode_intra_mb: s->current_picture.mb_type[mb_xy]= mb_type; if( cbp || IS_INTRA16x16( mb_type ) ) { - const uint8_t *scan, *scan8x8, *dc_scan; + const uint8_t *scan, *scan8x8; const uint32_t *qmul; if(IS_INTERLACED(mb_type)){ scan8x8= s->qscale ? h->field_scan8x8 : h->field_scan8x8_q0; scan= s->qscale ? h->field_scan : h->field_scan_q0; - dc_scan= luma_dc_field_scan; }else{ scan8x8= s->qscale ? h->zigzag_scan8x8 : h->zigzag_scan8x8_q0; scan= s->qscale ? h->zigzag_scan : h->zigzag_scan_q0; - dc_scan= luma_dc_zigzag_scan; } // decode_cabac_mb_dqp @@ -1642,7 +1640,9 @@ decode_intra_mb: if( IS_INTRA16x16( mb_type ) ) { int i; //av_log( s->avctx, AV_LOG_ERROR, "INTRA16x16 DC\n" ); - decode_cabac_residual_dc( h, h->mb, 0, 0, dc_scan, 16); + AV_ZERO128(h->mb_luma_dc+0); + AV_ZERO128(h->mb_luma_dc+8); + decode_cabac_residual_dc( h, h->mb_luma_dc, 0, 0, scan, 16); if( cbp&15 ) { qmul = h->dequant4_coeff[0][s->qscale]; |