MrBesen/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Move abs() from decode_cabac_mb_mvd() to the code that writes mvd_cache.

Browse Source 
4-8 cycles faster

Originally committed as revision 21887 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Michael Niedermayer 2010-02-18 23:37:48 +00:00
parent 8d63712486
commit a305449df6
1 changed files with 15 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
libavcodec/h264_cabac.c
View File

@ -910,8 +910,8 @@ static int decode_cabac_mb_ref( H264Context *h, int list, int n ) {
}
static int decode_cabac_mb_mvd( H264Context *h, int list, int n, int l ) {
int amvd = abs( h->mvd_cache[list][scan8[n] - 1][l] ) +
abs( h->mvd_cache[list][scan8[n] - 8][l] );
int amvd = h->mvd_cache[list][scan8[n] - 1][l] +
h->mvd_cache[list][scan8[n] - 8][l];
int ctxbase = (l == 0) ? 40 : 47;
int mvd;
int ctx = (amvd>2) + (amvd>32);
@ -1439,6 +1439,8 @@ decode_intra_mb:
my = mpy + decode_cabac_mb_mvd( h, list, index, 1 );
tprintf(s->avctx, "final mv:%d %d\n", mx, my);
mpx= abs(mpx-mx);
mpy= abs(mpy-my);
if(IS_SUB_8X8(sub_mb_type)){
mv_cache[ 1 ][0]=
mv_cache[ 8 ][0]= mv_cache[ 9 ][0]= mx;
@ -1446,27 +1448,27 @@ decode_intra_mb:
mv_cache[ 8 ][1]= mv_cache[ 9 ][1]= my;
mvd_cache[ 1 ][0]=
mvd_cache[ 8 ][0]= mvd_cache[ 9 ][0]= mx - mpx;
mvd_cache[ 8 ][0]= mvd_cache[ 9 ][0]= mpx;
mvd_cache[ 1 ][1]=
mvd_cache[ 8 ][1]= mvd_cache[ 9 ][1]= my - mpy;
mvd_cache[ 8 ][1]= mvd_cache[ 9 ][1]= mpy;
}else if(IS_SUB_8X4(sub_mb_type)){
mv_cache[ 1 ][0]= mx;
mv_cache[ 1 ][1]= my;
mvd_cache[ 1 ][0]= mx - mpx;
mvd_cache[ 1 ][1]= my - mpy;
mvd_cache[ 1 ][0]= mpx;
mvd_cache[ 1 ][1]= mpy;
}else if(IS_SUB_4X8(sub_mb_type)){
mv_cache[ 8 ][0]= mx;
mv_cache[ 8 ][1]= my;
mvd_cache[ 8 ][0]= mx - mpx;
mvd_cache[ 8 ][1]= my - mpy;
mvd_cache[ 8 ][0]= mpx;
mvd_cache[ 8 ][1]= mpy;
}
mv_cache[ 0 ][0]= mx;
mv_cache[ 0 ][1]= my;
mvd_cache[ 0 ][0]= mx - mpx;
mvd_cache[ 0 ][1]= my - mpy;
mvd_cache[ 0 ][0]= mpx;
mvd_cache[ 0 ][1]= mpy;
}
}else{
uint32_t *p= (uint32_t *)&h->mv_cache[list][ scan8[4*i] ][0];
@ -1507,7 +1509,7 @@ decode_intra_mb:
my = mpy + decode_cabac_mb_mvd( h, list, 0, 1 );
tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mvd_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(abs(mx-mpx),abs(my-mpy)), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx,my), 4);
}else
fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, 0, 4);
@ -1539,7 +1541,7 @@ decode_intra_mb:
my = mpy + decode_cabac_mb_mvd( h, list, 8*i, 1 );
tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(abs(mx-mpx),abs(my-mpy)), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx,my), 4);
}else{
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, 0, 4);
@ -1574,7 +1576,7 @@ decode_intra_mb:
my = mpy + decode_cabac_mb_mvd( h, list, 4*i, 1 );
tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(abs(mx-mpx),abs(my-mpy)), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx,my), 4);
}else{
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, 0, 4);