Mercurial > libavcodec.hg
diff vp8dsp.h @ 12238:1a7903913e9b libavcodec
VP8: 30% faster idct_mb
Take shortcuts based on statically common situations.
Add 4-at-a-time idct_dc function (mmx and sse2) since rows of 4 DC-only DCT
blocks are common.
TODO: tie this more directly into the MB mode, since the DC-level transform is
only used for non-splitmv blocks?
author | darkshikari |
---|---|
date | Fri, 23 Jul 2010 02:58:27 +0000 |
parents | 80b142c2e9f7 |
children | c7f6ddcc5c01 |
line wrap: on
line diff
--- a/vp8dsp.h Fri Jul 23 01:59:56 2010 +0000 +++ b/vp8dsp.h Fri Jul 23 02:58:27 2010 +0000 @@ -33,6 +33,7 @@ void (*vp8_luma_dc_wht)(DCTELEM block[4][4][16], DCTELEM dc[16]); void (*vp8_idct_add)(uint8_t *dst, DCTELEM block[16], int stride); void (*vp8_idct_dc_add)(uint8_t *dst, DCTELEM block[16], int stride); + void (*vp8_idct_dc_add4)(uint8_t *dst, DCTELEM block[4][16], int stride); // loop filter applied to edges between macroblocks void (*vp8_v_loop_filter16y)(uint8_t *dst, int stride,