diff options
| author | Ronald S. Bultje <rsbultje@gmail.com> | 2013-02-18 21:03:02 -0800 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2013-02-19 16:25:50 +0100 |
| commit | 1acd7d594c15aa491729c837ad3519d3469e620a (patch) | |
| tree | 4a515aa31c7483d8a2fdfa1d4399c9a9d2b29cae /libavcodec/h264pred_template.c | |
| parent | a1f1ca96b42698002920467c0aab9e636893088a (diff) | |
| download | ffmpeg-1acd7d594c15aa491729c837ad3519d3469e620a.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: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec/h264pred_template.c')
| -rw-r--r-- | libavcodec/h264pred_template.c | 28 |
1 files changed, 18 insertions, 10 deletions
diff --git a/libavcodec/h264pred_template.c b/libavcodec/h264pred_template.c index e78f2d423c..8d8d62e0b6 100644 --- a/libavcodec/h264pred_template.c +++ b/libavcodec/h264pred_template.c @@ -1132,7 +1132,7 @@ static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft, #undef PL #undef SRC -static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const int16_t *_block, +static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, int16_t *_block, ptrdiff_t stride) { int i; @@ -1149,9 +1149,11 @@ static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const int16_t *_block, pix++; block++; } + + memset(_block, 0, sizeof(dctcoef) * 16); } -static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const int16_t *_block, +static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, int16_t *_block, ptrdiff_t stride) { int i; @@ -1167,9 +1169,11 @@ static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const int16_t *_block, pix+= stride; block+= 4; } + + memset(_block, 0, sizeof(dctcoef) * 16); } -static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const int16_t *_block, +static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, int16_t *_block, ptrdiff_t stride) { int i; @@ -1190,9 +1194,11 @@ static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const int16_t *_block, pix++; block++; } + + memset(_block, 0, sizeof(dctcoef) * 64); } -static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const int16_t *_block, +static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, int16_t *_block, ptrdiff_t stride) { int i; @@ -1212,10 +1218,12 @@ static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const int16_t *_block, pix+= stride; block+= 8; } + + memset(_block, 0, sizeof(dctcoef) * 64); } static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, + int16_t *block, ptrdiff_t stride) { int i; @@ -1225,7 +1233,7 @@ static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset, static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, + int16_t *block, ptrdiff_t stride) { int i; @@ -1234,7 +1242,7 @@ static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, } static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, ptrdiff_t stride) + int16_t *block, ptrdiff_t stride) { int i; for(i=0; i<4; i++) @@ -1242,7 +1250,7 @@ static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, } static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, ptrdiff_t stride) + int16_t *block, ptrdiff_t stride) { int i; for(i=0; i<4; i++) @@ -1252,7 +1260,7 @@ static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset, } static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, + int16_t *block, ptrdiff_t stride) { int i; @@ -1262,7 +1270,7 @@ static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, static void FUNCC(pred8x16_horizontal_add)(uint8_t *pix, const int *block_offset, - const int16_t *block, ptrdiff_t stride) + int16_t *block, ptrdiff_t stride) { int i; for(i=0; i<4; i++) |