diff options
| author | Ronald S. Bultje <rsbultje@gmail.com> | 2013-02-18 21:03:02 -0800 |
|---|---|---|
| committer | Martin Storsjö <martin@martin.st> | 2013-04-10 11:03:06 +0300 |
| commit | 62844c3fd66940c7747e9b2bb7804e265319f43f (patch) | |
| tree | b0e5a05644457aa5d7598d1fefa1a41f83550753 /libavcodec/h264idct_template.c | |
| parent | e8cafd2773bc56455c8816593cbd9368f2d69a80 (diff) | |
| download | ffmpeg-62844c3fd66940c7747e9b2bb7804e265319f43f.tar.gz | |
h264: Integrate clear_blocks calls with IDCT
The non-intra-pcm branch in hl_decode_mb (simple, 8bpp) goes from 700
to 672 cycles, and the complete loop of decode_mb_cabac and hl_decode_mb
(in the decode_slice loop) goes from 1759 to 1733 cycles on the clip
tested (cathedral), i.e. almost 30 cycles per mb faster.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diffstat (limited to 'libavcodec/h264idct_template.c')
| -rw-r--r-- | libavcodec/h264idct_template.c | 16 |
1 files changed, 12 insertions, 4 deletions
diff --git a/libavcodec/h264idct_template.c b/libavcodec/h264idct_template.c index fc285c57b0..aadafaddd2 100644 --- a/libavcodec/h264idct_template.c +++ b/libavcodec/h264idct_template.c @@ -61,6 +61,8 @@ void FUNCC(ff_h264_idct_add)(uint8_t *_dst, int16_t *_block, int stride) dst[i + 2*stride]= av_clip_pixel(dst[i + 2*stride] + ((z1 - z2) >> 6)); dst[i + 3*stride]= av_clip_pixel(dst[i + 3*stride] + ((z0 - z3) >> 6)); } + + memset(block, 0, 16 * sizeof(dctcoef)); } void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride){ @@ -133,14 +135,18 @@ void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride){ dst[i + 6*stride] = av_clip_pixel( dst[i + 6*stride] + ((b2 - b5) >> 6) ); dst[i + 7*stride] = av_clip_pixel( dst[i + 7*stride] + ((b0 - b7) >> 6) ); } + + memset(block, 0, 64 * sizeof(dctcoef)); } // assumes all AC coefs are 0 -void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *block, int stride){ +void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *_block, int stride){ int i, j; - int dc = (((dctcoef*)block)[0] + 32) >> 6; pixel *dst = (pixel*)_dst; + dctcoef *block = (dctcoef*)_block; + int dc = (block[0] + 32) >> 6; stride /= sizeof(pixel); + block[0] = 0; for( j = 0; j < 4; j++ ) { for( i = 0; i < 4; i++ ) @@ -149,10 +155,12 @@ void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *block, int stride){ } } -void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, int16_t *block, int stride){ +void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, int16_t *_block, int stride){ int i, j; - int dc = (((dctcoef*)block)[0] + 32) >> 6; pixel *dst = (pixel*)_dst; + dctcoef *block = (dctcoef*)_block; + int dc = (block[0] + 32) >> 6; + block[0] = 0; stride /= sizeof(pixel); for( j = 0; j < 8; j++ ) { |