H.264/SVQ3: make chroma DC work the same way as luma DC

No speed improvement, but necessary for some future stuff.
Also opens up the possibility of asm chroma dc idct/dequant.

Originally committed as revision 26349 to svn://svn.ffmpeg.org/ffmpeg/trunk

1 files changed, 6 insertions, 91 deletions

diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 40dc276dd6..d0ee4744c8 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -246,93 +246,6 @@ int ff_h264_decode_rbsp_trailing(H264Context *h, const uint8_t *src){
     return 0;
 }
 
-#if 0
-/**
- * DCT transforms the 16 dc values.
- * @param qp quantization parameter ??? FIXME
- */
-static void h264_luma_dc_dct_c(DCTELEM *block/*, int qp*/){
-//    const int qmul= dequant_coeff[qp][0];
-    int i;
-    int temp[16]; //FIXME check if this is a good idea
-    static const int x_offset[4]={0, 1*stride, 4* stride,  5*stride};
-    static const int y_offset[4]={0, 2*stride, 8* stride, 10*stride};
-
-    for(i=0; i<4; i++){
-        const int offset= y_offset[i];
-        const int z0= block[offset+stride*0] + block[offset+stride*4];
-        const int z1= block[offset+stride*0] - block[offset+stride*4];
-        const int z2= block[offset+stride*1] - block[offset+stride*5];
-        const int z3= block[offset+stride*1] + block[offset+stride*5];
-
-        temp[4*i+0]= z0+z3;
-        temp[4*i+1]= z1+z2;
-        temp[4*i+2]= z1-z2;
-        temp[4*i+3]= z0-z3;
-    }
-
-    for(i=0; i<4; i++){
-        const int offset= x_offset[i];
-        const int z0= temp[4*0+i] + temp[4*2+i];
-        const int z1= temp[4*0+i] - temp[4*2+i];
-        const int z2= temp[4*1+i] - temp[4*3+i];
-        const int z3= temp[4*1+i] + temp[4*3+i];
-
-        block[stride*0 +offset]= (z0 + z3)>>1;
-        block[stride*2 +offset]= (z1 + z2)>>1;
-        block[stride*8 +offset]= (z1 - z2)>>1;
-        block[stride*10+offset]= (z0 - z3)>>1;
-    }
-}
-#endif
-
-#undef xStride
-#undef stride
-
-static void chroma_dc_dequant_idct_c(DCTELEM *block, int qmul){
-    const int stride= 16*2;
-    const int xStride= 16;
-    int a,b,c,d,e;
-
-    a= block[stride*0 + xStride*0];
-    b= block[stride*0 + xStride*1];
-    c= block[stride*1 + xStride*0];
-    d= block[stride*1 + xStride*1];
-
-    e= a-b;
-    a= a+b;
-    b= c-d;
-    c= c+d;
-
-    block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
-    block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
-    block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
-    block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
-}
-
-#if 0
-static void chroma_dc_dct_c(DCTELEM *block){
-    const int stride= 16*2;
-    const int xStride= 16;
-    int a,b,c,d,e;
-
-    a= block[stride*0 + xStride*0];
-    b= block[stride*0 + xStride*1];
-    c= block[stride*1 + xStride*0];
-    d= block[stride*1 + xStride*1];
-
-    e= a-b;
-    a= a+b;
-    b= c-d;
-    c= c+d;
-
-    block[stride*0 + xStride*0]= (a+c);
-    block[stride*0 + xStride*1]= (e+b);
-    block[stride*1 + xStride*0]= (a-c);
-    block[stride*1 + xStride*1]= (e-b);
-}
-#endif
-
 static inline void mc_dir_part(H264Context *h, Picture *pic, int n, int square, int chroma_height, int delta, int list,
                            uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                            int src_x_offset, int src_y_offset,
@@ -1283,17 +1196,19 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){
                     }
                 }
             }else{
+                int chroma_qpu = h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0];
+                int chroma_qpv = h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0];
                 if(is_h264){
                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+0] ])
-                        chroma_dc_dequant_idct_c(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+0*16, &h->mb_chroma_dc[0], chroma_qpu );
                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+1] ])
-                        chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+4*16, &h->mb_chroma_dc[1], chroma_qpv );
                     h->h264dsp.h264_idct_add8(dest, block_offset,
                                               h->mb, uvlinesize,
                                               h->non_zero_count_cache);
                 }else{
-                    chroma_dc_dequant_idct_c(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
-                    chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+0*16, &h->mb_chroma_dc[0], chroma_qpu );
+                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+4*16, &h->mb_chroma_dc[1], chroma_qpv );
                     for(i=16; i<16+8; i++){
                         if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]){
                             uint8_t * const ptr= dest[(i&4)>>2] + block_offset[i];