/**
* @file
- * H.264 / AVC / MPEG4 part10 prediction functions.
+ * H.264 / AVC / MPEG-4 part10 prediction functions.
* @author Michael Niedermayer <michaelni@gmx.at>
*/
-#include "mathops.h"
-#include "dsputil.h"
-
-#define BIT_DEPTH 8
+#include "libavutil/intreadwrite.h"
-#define pixel uint8_t
-#define pixel4 uint32_t
-#define dctcoef DCTELEM
+#include "mathops.h"
-#define INIT_CLIP uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
-#define CLIP(a) cm[a]
-#define FUNC(a) a
-#define FUNCC(a) a ## _c
-#define PIXEL_SPLAT_X4(x) ((x)*0x01010101U)
-#define AV_WN4P AV_WN32
-#define AV_WN4PA AV_WN32A
+#include "bit_depth_template.c"
-static void FUNCC(pred4x4_vertical)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_vertical)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- const pixel4 a= ((pixel4*)(src-stride))[0];
- ((pixel4*)(src+0*stride))[0]= a;
- ((pixel4*)(src+1*stride))[0]= a;
- ((pixel4*)(src+2*stride))[0]= a;
- ((pixel4*)(src+3*stride))[0]= a;
+ const pixel4 a= AV_RN4PA(src-stride);
+
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_horizontal)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_horizontal)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- ((pixel4*)(src+0*stride))[0]= PIXEL_SPLAT_X4(src[-1+0*stride]);
- ((pixel4*)(src+1*stride))[0]= PIXEL_SPLAT_X4(src[-1+1*stride]);
- ((pixel4*)(src+2*stride))[0]= PIXEL_SPLAT_X4(src[-1+2*stride]);
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(src[-1+3*stride]);
+ AV_WN4PA(src+0*stride, PIXEL_SPLAT_X4(src[-1+0*stride]));
+ AV_WN4PA(src+1*stride, PIXEL_SPLAT_X4(src[-1+1*stride]));
+ AV_WN4PA(src+2*stride, PIXEL_SPLAT_X4(src[-1+2*stride]));
+ AV_WN4PA(src+3*stride, PIXEL_SPLAT_X4(src[-1+3*stride]));
}
-static void FUNCC(pred4x4_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
+ src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
+ const pixel4 a = PIXEL_SPLAT_X4(dc);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_left_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_left_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
+ const pixel4 a = PIXEL_SPLAT_X4(dc);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_top_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_top_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
+ const pixel4 a = PIXEL_SPLAT_X4(dc);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(dc);
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_128_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_128_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1));
+ const pixel4 a = PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1));
+
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_127_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_127_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))-1);
+ const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))-1);
+
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
-static void FUNCC(pred4x4_129_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_129_dc)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- ((pixel4*)(src+0*stride))[0]=
- ((pixel4*)(src+1*stride))[0]=
- ((pixel4*)(src+2*stride))[0]=
- ((pixel4*)(src+3*stride))[0]= PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))+1);
+ const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))+1);
+
+ AV_WN4PA(src+0*stride, a);
+ AV_WN4PA(src+1*stride, a);
+ AV_WN4PA(src+2*stride, a);
+ AV_WN4PA(src+3*stride, a);
}
#define LOAD_TOP_RIGHT_EDGE\
- const int av_unused t4= topright[0];\
- const int av_unused t5= topright[1];\
- const int av_unused t6= topright[2];\
- const int av_unused t7= topright[3];\
+ const unsigned av_unused t4 = topright[0];\
+ const unsigned av_unused t5 = topright[1];\
+ const unsigned av_unused t6 = topright[2];\
+ const unsigned av_unused t7 = topright[3];\
#define LOAD_DOWN_LEFT_EDGE\
- const int av_unused l4= src[-1+4*stride];\
- const int av_unused l5= src[-1+5*stride];\
- const int av_unused l6= src[-1+6*stride];\
- const int av_unused l7= src[-1+7*stride];\
+ const unsigned av_unused l4 = src[-1+4*stride];\
+ const unsigned av_unused l5 = src[-1+5*stride];\
+ const unsigned av_unused l6 = src[-1+6*stride];\
+ const unsigned av_unused l7 = src[-1+7*stride];\
#define LOAD_LEFT_EDGE\
- const int av_unused l0= src[-1+0*stride];\
- const int av_unused l1= src[-1+1*stride];\
- const int av_unused l2= src[-1+2*stride];\
- const int av_unused l3= src[-1+3*stride];\
+ const unsigned av_unused l0 = src[-1+0*stride];\
+ const unsigned av_unused l1 = src[-1+1*stride];\
+ const unsigned av_unused l2 = src[-1+2*stride];\
+ const unsigned av_unused l3 = src[-1+3*stride];\
#define LOAD_TOP_EDGE\
- const int av_unused t0= src[ 0-1*stride];\
- const int av_unused t1= src[ 1-1*stride];\
- const int av_unused t2= src[ 2-1*stride];\
- const int av_unused t3= src[ 3-1*stride];\
+ const unsigned av_unused t0 = src[ 0-1*stride];\
+ const unsigned av_unused t1 = src[ 1-1*stride];\
+ const unsigned av_unused t2 = src[ 2-1*stride];\
+ const unsigned av_unused t3 = src[ 3-1*stride];\
-static void FUNCC(pred4x4_down_right)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_down_right)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int lt= src[-1-1*stride];
src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
}
-static void FUNCC(pred4x4_down_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
+static void FUNCC(pred4x4_down_left)(uint8_t *_src, const uint8_t *_topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
const pixel *topright = (const pixel*)_topright;
int stride = _stride/sizeof(pixel);
src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
}
-static void FUNCC(pred4x4_vertical_right)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_vertical_right)(uint8_t *_src,
+ const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int lt= src[-1-1*stride];
src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
}
-static void FUNCC(pred4x4_vertical_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
+static void FUNCC(pred4x4_vertical_left)(uint8_t *_src,
+ const uint8_t *_topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
const pixel *topright = (const pixel*)_topright;
int stride = _stride/sizeof(pixel);
src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
}
-static void FUNCC(pred4x4_horizontal_up)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_horizontal_up)(uint8_t *_src, const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
LOAD_LEFT_EDGE
src[3+3*stride]=l3;
}
-static void FUNCC(pred4x4_horizontal_down)(uint8_t *_src, const uint8_t *topright, int _stride){
+static void FUNCC(pred4x4_horizontal_down)(uint8_t *_src,
+ const uint8_t *topright,
+ ptrdiff_t _stride)
+{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const int lt= src[-1-1*stride];
src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
}
-static void FUNCC(pred16x16_vertical)(uint8_t *_src, int _stride){
+static void FUNCC(pred16x16_vertical)(uint8_t *_src, ptrdiff_t _stride)
+{
int i;
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- const pixel4 a = ((pixel4*)(src-stride))[0];
- const pixel4 b = ((pixel4*)(src-stride))[1];
- const pixel4 c = ((pixel4*)(src-stride))[2];
- const pixel4 d = ((pixel4*)(src-stride))[3];
+ const pixel4 a = AV_RN4PA(((pixel4*)(src-stride))+0);
+ const pixel4 b = AV_RN4PA(((pixel4*)(src-stride))+1);
+ const pixel4 c = AV_RN4PA(((pixel4*)(src-stride))+2);
+ const pixel4 d = AV_RN4PA(((pixel4*)(src-stride))+3);
for(i=0; i<16; i++){
- ((pixel4*)(src+i*stride))[0] = a;
- ((pixel4*)(src+i*stride))[1] = b;
- ((pixel4*)(src+i*stride))[2] = c;
- ((pixel4*)(src+i*stride))[3] = d;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
+ AV_WN4PA(((pixel4*)(src+i*stride))+2, c);
+ AV_WN4PA(((pixel4*)(src+i*stride))+3, d);
}
}
-static void FUNCC(pred16x16_horizontal)(uint8_t *_src, int stride){
+static void FUNCC(pred16x16_horizontal)(uint8_t *_src, ptrdiff_t stride)
+{
int i;
pixel *src = (pixel*)_src;
stride /= sizeof(pixel);
for(i=0; i<16; i++){
- ((pixel4*)(src+i*stride))[0] =
- ((pixel4*)(src+i*stride))[1] =
- ((pixel4*)(src+i*stride))[2] =
- ((pixel4*)(src+i*stride))[3] = PIXEL_SPLAT_X4(src[-1+i*stride]);
+ const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
+
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+2, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+3, a);
}
}
#define PREDICT_16x16_DC(v)\
for(i=0; i<16; i++){\
- AV_WN4P(src+ 0, v);\
- AV_WN4P(src+ 4, v);\
- AV_WN4P(src+ 8, v);\
- AV_WN4P(src+12, v);\
+ AV_WN4PA(src+ 0, v);\
+ AV_WN4PA(src+ 4, v);\
+ AV_WN4PA(src+ 8, v);\
+ AV_WN4PA(src+12, v);\
src += stride;\
}
-static void FUNCC(pred16x16_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred16x16_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i, dc=0;
pixel *src = (pixel*)_src;
pixel4 dcsplat;
PREDICT_16x16_DC(dcsplat);
}
-static void FUNCC(pred16x16_left_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred16x16_left_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i, dc=0;
pixel *src = (pixel*)_src;
pixel4 dcsplat;
PREDICT_16x16_DC(dcsplat);
}
-static void FUNCC(pred16x16_top_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred16x16_top_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i, dc=0;
pixel *src = (pixel*)_src;
pixel4 dcsplat;
}
#define PRED16x16_X(n, v) \
-static void FUNCC(pred16x16_##n##_dc)(uint8_t *_src, int stride){\
+static void FUNCC(pred16x16_##n##_dc)(uint8_t *_src, ptrdiff_t stride)\
+{\
int i;\
pixel *src = (pixel*)_src;\
stride /= sizeof(pixel);\
PREDICT_16x16_DC(PIXEL_SPLAT_X4(v));\
}
-PRED16x16_X(127, (1<<(BIT_DEPTH-1))-1);
-PRED16x16_X(128, (1<<(BIT_DEPTH-1))+0);
-PRED16x16_X(129, (1<<(BIT_DEPTH-1))+1);
+PRED16x16_X(127, (1<<(BIT_DEPTH-1))-1)
+PRED16x16_X(128, (1<<(BIT_DEPTH-1))+0)
+PRED16x16_X(129, (1<<(BIT_DEPTH-1))+1)
-static inline void FUNCC(pred16x16_plane_compat)(uint8_t *_src, int _stride, const int svq3, const int rv40){
+static inline void FUNCC(pred16x16_plane_compat)(uint8_t *_src,
+ ptrdiff_t _stride,
+ const int svq3,
+ const int rv40)
+{
int i, j, k;
int a;
INIT_CLIP
}
}
-static void FUNCC(pred16x16_plane)(uint8_t *src, int stride){
+static void FUNCC(pred16x16_plane)(uint8_t *src, ptrdiff_t stride)
+{
FUNCC(pred16x16_plane_compat)(src, stride, 0, 0);
}
-static void FUNCC(pred8x8_vertical)(uint8_t *_src, int _stride){
+static void FUNCC(pred8x8_vertical)(uint8_t *_src, ptrdiff_t _stride)
+{
int i;
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
- const pixel4 a= ((pixel4*)(src-stride))[0];
- const pixel4 b= ((pixel4*)(src-stride))[1];
+ const pixel4 a= AV_RN4PA(((pixel4*)(src-stride))+0);
+ const pixel4 b= AV_RN4PA(((pixel4*)(src-stride))+1);
for(i=0; i<8; i++){
- ((pixel4*)(src+i*stride))[0]= a;
- ((pixel4*)(src+i*stride))[1]= b;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
+ }
+}
+
+static void FUNCC(pred8x16_vertical)(uint8_t *_src, ptrdiff_t _stride)
+{
+ int i;
+ pixel *src = (pixel*)_src;
+ int stride = _stride>>(sizeof(pixel)-1);
+ const pixel4 a= AV_RN4PA(((pixel4*)(src-stride))+0);
+ const pixel4 b= AV_RN4PA(((pixel4*)(src-stride))+1);
+
+ for(i=0; i<16; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
}
}
-static void FUNCC(pred8x8_horizontal)(uint8_t *_src, int stride){
+static void FUNCC(pred8x8_horizontal)(uint8_t *_src, ptrdiff_t stride)
+{
int i;
pixel *src = (pixel*)_src;
stride /= sizeof(pixel);
for(i=0; i<8; i++){
- ((pixel4*)(src+i*stride))[0]=
- ((pixel4*)(src+i*stride))[1]= PIXEL_SPLAT_X4(src[-1+i*stride]);
+ const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
+ }
+}
+
+static void FUNCC(pred8x16_horizontal)(uint8_t *_src, ptrdiff_t stride)
+{
+ int i;
+ pixel *src = (pixel*)_src;
+ stride >>= sizeof(pixel)-1;
+ for(i=0; i<16; i++){
+ const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
}
}
#define PRED8x8_X(n, v)\
-static void FUNCC(pred8x8_##n##_dc)(uint8_t *_src, int stride){\
+static void FUNCC(pred8x8_##n##_dc)(uint8_t *_src, ptrdiff_t stride)\
+{\
int i;\
+ const pixel4 a = PIXEL_SPLAT_X4(v);\
pixel *src = (pixel*)_src;\
stride /= sizeof(pixel);\
for(i=0; i<8; i++){\
- ((pixel4*)(src+i*stride))[0]=\
- ((pixel4*)(src+i*stride))[1]= PIXEL_SPLAT_X4(v);\
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, a);\
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, a);\
}\
}
-PRED8x8_X(127, (1<<(BIT_DEPTH-1))-1);
-PRED8x8_X(128, (1<<(BIT_DEPTH-1))+0);
-PRED8x8_X(129, (1<<(BIT_DEPTH-1))+1);
+PRED8x8_X(127, (1<<(BIT_DEPTH-1))-1)
+PRED8x8_X(128, (1<<(BIT_DEPTH-1))+0)
+PRED8x8_X(129, (1<<(BIT_DEPTH-1))+1)
+
+static void FUNCC(pred8x16_128_dc)(uint8_t *_src, ptrdiff_t stride)
+{
+ FUNCC(pred8x8_128_dc)(_src, stride);
+ FUNCC(pred8x8_128_dc)(_src+8*stride, stride);
+}
-static void FUNCC(pred8x8_left_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred8x8_left_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i;
int dc0, dc2;
pixel4 dc0splat, dc2splat;
dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
for(i=0; i<4; i++){
- ((pixel4*)(src+i*stride))[0]=
- ((pixel4*)(src+i*stride))[1]= dc0splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc0splat);
}
for(i=4; i<8; i++){
- ((pixel4*)(src+i*stride))[0]=
- ((pixel4*)(src+i*stride))[1]= dc2splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc2splat);
}
}
-static void FUNCC(pred8x8_top_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred8x16_left_dc)(uint8_t *_src, ptrdiff_t stride)
+{
+ FUNCC(pred8x8_left_dc)(_src, stride);
+ FUNCC(pred8x8_left_dc)(_src+8*stride, stride);
+}
+
+static void FUNCC(pred8x8_top_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i;
int dc0, dc1;
pixel4 dc0splat, dc1splat;
dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
for(i=0; i<4; i++){
- ((pixel4*)(src+i*stride))[0]= dc0splat;
- ((pixel4*)(src+i*stride))[1]= dc1splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
}
for(i=4; i<8; i++){
- ((pixel4*)(src+i*stride))[0]= dc0splat;
- ((pixel4*)(src+i*stride))[1]= dc1splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
}
}
-static void FUNCC(pred8x8_dc)(uint8_t *_src, int stride){
+static void FUNCC(pred8x16_top_dc)(uint8_t *_src, ptrdiff_t stride)
+{
+ int i;
+ int dc0, dc1;
+ pixel4 dc0splat, dc1splat;
+ pixel *src = (pixel*)_src;
+ stride >>= sizeof(pixel)-1;
+
+ dc0=dc1=0;
+ for(i=0;i<4; i++){
+ dc0+= src[i-stride];
+ dc1+= src[4+i-stride];
+ }
+ dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
+ dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
+
+ for(i=0; i<16; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
+ }
+}
+
+static void FUNCC(pred8x8_dc)(uint8_t *_src, ptrdiff_t stride)
+{
int i;
int dc0, dc1, dc2;
pixel4 dc0splat, dc1splat, dc2splat, dc3splat;
dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);
for(i=0; i<4; i++){
- ((pixel4*)(src+i*stride))[0]= dc0splat;
- ((pixel4*)(src+i*stride))[1]= dc1splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
}
for(i=4; i<8; i++){
- ((pixel4*)(src+i*stride))[0]= dc2splat;
- ((pixel4*)(src+i*stride))[1]= dc3splat;
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc3splat);
}
}
-//the following 4 function should not be optimized!
-static void FUNC(pred8x8_mad_cow_dc_l0t)(uint8_t *src, int stride){
+static void FUNCC(pred8x16_dc)(uint8_t *_src, ptrdiff_t stride)
+{
+ int i;
+ int dc0, dc1, dc2, dc3, dc4;
+ pixel4 dc0splat, dc1splat, dc2splat, dc3splat, dc4splat, dc5splat, dc6splat, dc7splat;
+ pixel *src = (pixel*)_src;
+ stride >>= sizeof(pixel)-1;
+
+ dc0=dc1=dc2=dc3=dc4=0;
+ for(i=0;i<4; i++){
+ dc0+= src[-1+i*stride] + src[i-stride];
+ dc1+= src[4+i-stride];
+ dc2+= src[-1+(i+4)*stride];
+ dc3+= src[-1+(i+8)*stride];
+ dc4+= src[-1+(i+12)*stride];
+ }
+ dc0splat = PIXEL_SPLAT_X4((dc0 + 4)>>3);
+ dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
+ dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
+ dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);
+ dc4splat = PIXEL_SPLAT_X4((dc3 + 2)>>2);
+ dc5splat = PIXEL_SPLAT_X4((dc1 + dc3 + 4)>>3);
+ dc6splat = PIXEL_SPLAT_X4((dc4 + 2)>>2);
+ dc7splat = PIXEL_SPLAT_X4((dc1 + dc4 + 4)>>3);
+
+ for(i=0; i<4; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
+ }
+ for(i=4; i<8; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc3splat);
+ }
+ for(i=8; i<12; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc4splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc5splat);
+ }
+ for(i=12; i<16; i++){
+ AV_WN4PA(((pixel4*)(src+i*stride))+0, dc6splat);
+ AV_WN4PA(((pixel4*)(src+i*stride))+1, dc7splat);
+ }
+}
+
+static void FUNC(pred8x8_mad_cow_dc_l0t)(uint8_t *src, ptrdiff_t stride)
+{
FUNCC(pred8x8_top_dc)(src, stride);
FUNCC(pred4x4_dc)(src, NULL, stride);
}
-static void FUNC(pred8x8_mad_cow_dc_0lt)(uint8_t *src, int stride){
+static void FUNC(pred8x16_mad_cow_dc_l0t)(uint8_t *src, ptrdiff_t stride)
+{
+ FUNCC(pred8x16_top_dc)(src, stride);
+ FUNCC(pred4x4_dc)(src, NULL, stride);
+}
+
+static void FUNC(pred8x8_mad_cow_dc_0lt)(uint8_t *src, ptrdiff_t stride)
+{
FUNCC(pred8x8_dc)(src, stride);
FUNCC(pred4x4_top_dc)(src, NULL, stride);
}
-static void FUNC(pred8x8_mad_cow_dc_l00)(uint8_t *src, int stride){
+static void FUNC(pred8x16_mad_cow_dc_0lt)(uint8_t *src, ptrdiff_t stride)
+{
+ FUNCC(pred8x16_dc)(src, stride);
+ FUNCC(pred4x4_top_dc)(src, NULL, stride);
+}
+
+static void FUNC(pred8x8_mad_cow_dc_l00)(uint8_t *src, ptrdiff_t stride)
+{
FUNCC(pred8x8_left_dc)(src, stride);
FUNCC(pred4x4_128_dc)(src + 4*stride , NULL, stride);
FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);
}
-static void FUNC(pred8x8_mad_cow_dc_0l0)(uint8_t *src, int stride){
+static void FUNC(pred8x16_mad_cow_dc_l00)(uint8_t *src, ptrdiff_t stride)
+{
+ FUNCC(pred8x16_left_dc)(src, stride);
+ FUNCC(pred4x4_128_dc)(src + 4*stride , NULL, stride);
+ FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);
+}
+
+static void FUNC(pred8x8_mad_cow_dc_0l0)(uint8_t *src, ptrdiff_t stride)
+{
FUNCC(pred8x8_left_dc)(src, stride);
FUNCC(pred4x4_128_dc)(src , NULL, stride);
FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);
}
-static void FUNCC(pred8x8_plane)(uint8_t *_src, int _stride){
+static void FUNC(pred8x16_mad_cow_dc_0l0)(uint8_t *src, ptrdiff_t stride)
+{
+ FUNCC(pred8x16_left_dc)(src, stride);
+ FUNCC(pred4x4_128_dc)(src , NULL, stride);
+ FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);
+}
+
+static void FUNCC(pred8x8_plane)(uint8_t *_src, ptrdiff_t _stride)
+{
int j, k;
int a;
INIT_CLIP
}
}
+static void FUNCC(pred8x16_plane)(uint8_t *_src, ptrdiff_t _stride)
+{
+ int j, k;
+ int a;
+ INIT_CLIP
+ pixel *src = (pixel*)_src;
+ int stride = _stride>>(sizeof(pixel)-1);
+ const pixel * const src0 = src +3-stride;
+ const pixel * src1 = src +8*stride-1;
+ const pixel * src2 = src1-2*stride; // == src+6*stride-1;
+ int H = src0[1] - src0[-1];
+ int V = src1[0] - src2[ 0];
+
+ for (k = 2; k <= 4; ++k) {
+ src1 += stride; src2 -= stride;
+ H += k*(src0[k] - src0[-k]);
+ V += k*(src1[0] - src2[ 0]);
+ }
+ for (; k <= 8; ++k) {
+ src1 += stride; src2 -= stride;
+ V += k*(src1[0] - src2[0]);
+ }
+
+ H = (17*H+16) >> 5;
+ V = (5*V+32) >> 6;
+
+ a = 16*(src1[0] + src2[8] + 1) - 7*V - 3*H;
+ for(j=16; j>0; --j) {
+ int b = a;
+ a += V;
+ src[0] = CLIP((b ) >> 5);
+ src[1] = CLIP((b+ H) >> 5);
+ src[2] = CLIP((b+2*H) >> 5);
+ src[3] = CLIP((b+3*H) >> 5);
+ src[4] = CLIP((b+4*H) >> 5);
+ src[5] = CLIP((b+5*H) >> 5);
+ src[6] = CLIP((b+6*H) >> 5);
+ src[7] = CLIP((b+7*H) >> 5);
+ src += stride;
+ }
+}
+
#define SRC(x,y) src[(x)+(y)*stride]
#define PL(y) \
const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
#define PREDICT_8x8_DC(v) \
int y; \
for( y = 0; y < 8; y++ ) { \
- ((pixel4*)src)[0] = \
- ((pixel4*)src)[1] = v; \
+ AV_WN4PA(((pixel4*)src)+0, v); \
+ AV_WN4PA(((pixel4*)src)+1, v); \
src += stride; \
}
-static void FUNCC(pred8x8l_128_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_128_dc)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
PREDICT_8x8_DC(PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1)));
}
-static void FUNCC(pred8x8l_left_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_left_dc)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3);
PREDICT_8x8_DC(dc);
}
-static void FUNCC(pred8x8l_top_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_top_dc)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
const pixel4 dc = PIXEL_SPLAT_X4((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3);
PREDICT_8x8_DC(dc);
}
-static void FUNCC(pred8x8l_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_dc)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
+t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4);
PREDICT_8x8_DC(dc);
}
-static void FUNCC(pred8x8l_horizontal)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_horizontal)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
+ pixel4 a;
PREDICT_8x8_LOAD_LEFT;
-#define ROW(y) ((pixel4*)(src+y*stride))[0] =\
- ((pixel4*)(src+y*stride))[1] = PIXEL_SPLAT_X4(l##y)
+#define ROW(y) a = PIXEL_SPLAT_X4(l##y); \
+ AV_WN4PA(src+y*stride, a); \
+ AV_WN4PA(src+y*stride+4, a);
ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
#undef ROW
}
-static void FUNCC(pred8x8l_vertical)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_vertical)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
int y;
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
+ pixel4 a, b;
PREDICT_8x8_LOAD_TOP;
src[0] = t0;
src[5] = t5;
src[6] = t6;
src[7] = t7;
+ a = AV_RN4PA(((pixel4*)src)+0);
+ b = AV_RN4PA(((pixel4*)src)+1);
for( y = 1; y < 8; y++ ) {
- ((pixel4*)(src+y*stride))[0] = ((pixel4*)src)[0];
- ((pixel4*)(src+y*stride))[1] = ((pixel4*)src)[1];
+ AV_WN4PA(((pixel4*)(src+y*stride))+0, a);
+ AV_WN4PA(((pixel4*)(src+y*stride))+1, b);
}
}
-static void FUNCC(pred8x8l_down_left)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_down_left)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
}
-static void FUNCC(pred8x8l_down_right)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_down_right)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
}
-static void FUNCC(pred8x8l_vertical_right)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_vertical_right)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
SRC(7,0)= (t6 + t7 + 1) >> 1;
}
-static void FUNCC(pred8x8l_horizontal_down)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_horizontal_down)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
}
-static void FUNCC(pred8x8l_vertical_left)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_vertical_left)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(7,6)= (t10 + t11 + 1) >> 1;
SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
}
-static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
+static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
{
pixel *src = (pixel*)_src;
int stride = _stride/sizeof(pixel);
SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
}
+
+static void FUNCC(pred8x8l_vertical_filter_add)(uint8_t *_src, int16_t *_block, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
+{
+ int i;
+ pixel *src = (pixel*)_src;
+ const dctcoef *block = (const dctcoef*)_block;
+ pixel pix[8];
+ int stride = _stride/sizeof(pixel);
+ PREDICT_8x8_LOAD_TOP;
+
+ pix[0] = t0;
+ pix[1] = t1;
+ pix[2] = t2;
+ pix[3] = t3;
+ pix[4] = t4;
+ pix[5] = t5;
+ pix[6] = t6;
+ pix[7] = t7;
+
+ for (i = 0; i < 8; i++) {
+ pixel v = pix[i];
+ src[0 * stride] = v += block[0];
+ src[1 * stride] = v += block[8];
+ src[2 * stride] = v += block[16];
+ src[3 * stride] = v += block[24];
+ src[4 * stride] = v += block[32];
+ src[5 * stride] = v += block[40];
+ src[6 * stride] = v += block[48];
+ src[7 * stride] = v + block[56];
+ src++;
+ block++;
+ }
+
+ memset(_block, 0, sizeof(dctcoef) * 64);
+}
+
+static void FUNCC(pred8x8l_horizontal_filter_add)(uint8_t *_src, int16_t *_block, int has_topleft,
+ int has_topright, ptrdiff_t _stride)
+{
+ int i;
+ pixel *src = (pixel*)_src;
+ const dctcoef *block = (const dctcoef*)_block;
+ pixel pix[8];
+ int stride = _stride/sizeof(pixel);
+ PREDICT_8x8_LOAD_LEFT;
+
+ pix[0] = l0;
+ pix[1] = l1;
+ pix[2] = l2;
+ pix[3] = l3;
+ pix[4] = l4;
+ pix[5] = l5;
+ pix[6] = l6;
+ pix[7] = l7;
+
+ for (i = 0; i < 8; i++) {
+ pixel v = pix[i];
+ src[0] = v += block[0];
+ src[1] = v += block[1];
+ src[2] = v += block[2];
+ src[3] = v += block[3];
+ src[4] = v += block[4];
+ src[5] = v += block[5];
+ src[6] = v += block[6];
+ src[7] = v + block[7];
+ src += stride;
+ block += 8;
+ }
+
+ memset(_block, 0, sizeof(dctcoef) * 64);
+}
+
#undef PREDICT_8x8_LOAD_LEFT
#undef PREDICT_8x8_LOAD_TOP
#undef PREDICT_8x8_LOAD_TOPLEFT
#undef PL
#undef SRC
-static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const DCTELEM *_block, int stride){
+static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, int16_t *_block,
+ ptrdiff_t stride)
+{
int i;
pixel *pix = (pixel*)_pix;
const dctcoef *block = (const dctcoef*)_block;
pix++;
block++;
}
+
+ memset(_block, 0, sizeof(dctcoef) * 16);
}
-static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const DCTELEM *_block, int stride){
+static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, int16_t *_block,
+ ptrdiff_t stride)
+{
int i;
pixel *pix = (pixel*)_pix;
const dctcoef *block = (const dctcoef*)_block;
pix+= stride;
block+= 4;
}
+
+ memset(_block, 0, sizeof(dctcoef) * 16);
}
-static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const DCTELEM *_block, int stride){
+static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, int16_t *_block,
+ ptrdiff_t stride)
+{
int i;
pixel *pix = (pixel*)_pix;
const dctcoef *block = (const dctcoef*)_block;
pix++;
block++;
}
+
+ memset(_block, 0, sizeof(dctcoef) * 64);
}
-static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const DCTELEM *_block, int stride){
+static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, int16_t *_block,
+ ptrdiff_t stride)
+{
int i;
pixel *pix = (pixel*)_pix;
const dctcoef *block = (const dctcoef*)_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 DCTELEM *block, int stride){
+static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset,
+ int16_t *block,
+ ptrdiff_t stride)
+{
int i;
for(i=0; i<16; i++)
FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
}
-static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
+static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix,
+ const int *block_offset,
+ int16_t *block,
+ ptrdiff_t stride)
+{
int i;
for(i=0; i<16; i++)
FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
}
-static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
+static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset,
+ int16_t *block, ptrdiff_t stride)
+{
+ int i;
+ for(i=0; i<4; i++)
+ FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
+}
+
+static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset,
+ int16_t *block, ptrdiff_t stride)
+{
int i;
for(i=0; i<4; i++)
FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
+ for(i=4; i<8; i++)
+ FUNCC(pred4x4_vertical_add)(pix + block_offset[i+4], block + i*16*sizeof(pixel), stride);
+}
+
+static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset,
+ int16_t *block,
+ ptrdiff_t stride)
+{
+ int i;
+ for(i=0; i<4; i++)
+ FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
}
-static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
+static void FUNCC(pred8x16_horizontal_add)(uint8_t *pix,
+ const int *block_offset,
+ int16_t *block, ptrdiff_t stride)
+{
int i;
for(i=0; i<4; i++)
FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
+ for(i=4; i<8; i++)
+ FUNCC(pred4x4_horizontal_add)(pix + block_offset[i+4], block + i*16*sizeof(pixel), stride);
}