*/
#include "avcodec.h"
#include "dsputil.h"
-
-void (*get_pixels)(DCTELEM *block, const UINT8 *pixels, int line_size);
-void (*diff_pixels)(DCTELEM *block, const UINT8 *s1, const UINT8 *s2, int stride);
-void (*put_pixels_clamped)(const DCTELEM *block, UINT8 *pixels, int line_size);
-void (*add_pixels_clamped)(const DCTELEM *block, UINT8 *pixels, int line_size);
-void (*gmc1)(UINT8 *dst, UINT8 *src, int srcStride, int h, int x16, int y16, int rounder);
-void (*clear_blocks)(DCTELEM *blocks);
-int (*pix_sum)(UINT8 * pix, int line_size);
-int (*pix_norm1)(UINT8 * pix, int line_size);
-
-op_pixels_abs_func pix_abs16x16;
-op_pixels_abs_func pix_abs16x16_x2;
-op_pixels_abs_func pix_abs16x16_y2;
-op_pixels_abs_func pix_abs16x16_xy2;
-
-op_pixels_abs_func pix_abs8x8;
-op_pixels_abs_func pix_abs8x8_x2;
-op_pixels_abs_func pix_abs8x8_y2;
-op_pixels_abs_func pix_abs8x8_xy2;
+#include "mpegvideo.h"
int ff_bit_exact=0;
};
/* a*inverse[b]>>32 == a/b for all 0<=a<=65536 && 2<=b<=255 */
-UINT32 inverse[256]={
+const UINT32 inverse[256]={
0, 4294967295U,2147483648U,1431655766, 1073741824, 858993460, 715827883, 613566757,
536870912, 477218589, 429496730, 390451573, 357913942, 330382100, 306783379, 286331154,
268435456, 252645136, 238609295, 226050911, 214748365, 204522253, 195225787, 186737709,
17318417, 17248865, 17179870, 17111424, 17043522, 16976156, 16909321, 16843010,
};
-int pix_sum_c(UINT8 * pix, int line_size)
+static int pix_sum_c(UINT8 * pix, int line_size)
{
int s, i, j;
return s;
}
-int pix_norm1_c(UINT8 * pix, int line_size)
+static int pix_norm1_c(UINT8 * pix, int line_size)
{
int s, i, j;
UINT32 *sq = squareTbl + 256;
}
-void get_pixels_c(DCTELEM *restrict block, const UINT8 *pixels, int line_size)
+static int sse8_c(void *v, UINT8 * pix1, UINT8 * pix2, int line_size)
+{
+ int s, i;
+ UINT32 *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < 8; i++) {
+ s += sq[pix1[0] - pix2[0]];
+ s += sq[pix1[1] - pix2[1]];
+ s += sq[pix1[2] - pix2[2]];
+ s += sq[pix1[3] - pix2[3]];
+ s += sq[pix1[4] - pix2[4]];
+ s += sq[pix1[5] - pix2[5]];
+ s += sq[pix1[6] - pix2[6]];
+ s += sq[pix1[7] - pix2[7]];
+ pix1 += line_size;
+ pix2 += line_size;
+ }
+ return s;
+}
+
+static int sse16_c(void *v, UINT8 * pix1, UINT8 * pix2, int line_size)
+{
+ int s, i, j;
+ UINT32 *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < 16; i++) {
+ for (j = 0; j < 16; j += 8) {
+ s += sq[pix1[0] - pix2[0]];
+ s += sq[pix1[1] - pix2[1]];
+ s += sq[pix1[2] - pix2[2]];
+ s += sq[pix1[3] - pix2[3]];
+ s += sq[pix1[4] - pix2[4]];
+ s += sq[pix1[5] - pix2[5]];
+ s += sq[pix1[6] - pix2[6]];
+ s += sq[pix1[7] - pix2[7]];
+ pix1 += 8;
+ pix2 += 8;
+ }
+ pix1 += line_size - 16;
+ pix2 += line_size - 16;
+ }
+ return s;
+}
+
+static void get_pixels_c(DCTELEM *restrict block, const UINT8 *pixels, int line_size)
{
int i;
}
}
-void diff_pixels_c(DCTELEM *restrict block, const UINT8 *s1, const UINT8 *s2,
- int stride){
+static void diff_pixels_c(DCTELEM *restrict block, const UINT8 *s1,
+ const UINT8 *s2, int stride){
int i;
/* read the pixels */
}
-void put_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
- int line_size)
+static void put_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
+ int line_size)
{
int i;
UINT8 *cm = cropTbl + MAX_NEG_CROP;
}
}
-void add_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
+static void add_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
int line_size)
{
int i;
}\
}\
\
-static void OPNAME ## _no_rnd_pixels_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _no_rnd_pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
for(i=0; i<h; i++){\
}\
}\
\
-static void OPNAME ## _pixels_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
for(i=0; i<h; i++){\
}\
}\
\
-static void OPNAME ## _no_rnd_pixels_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _no_rnd_pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
for(i=0; i<h; i++){\
}\
}\
\
-static void OPNAME ## _pixels_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
for(i=0; i<h; i++){\
}\
}\
\
-static void OPNAME ## _pixels_xy2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
const uint64_t a= LD64(pixels );\
}\
}\
\
-static void OPNAME ## _no_rnd_pixels_xy2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _no_rnd_pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
const uint64_t a= LD64(pixels );\
}\
}\
\
-CALL_2X_PIXELS(OPNAME ## _pixels16 , OPNAME ## _pixels , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_x2 , OPNAME ## _pixels_x2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_y2 , OPNAME ## _pixels_y2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_xy2, OPNAME ## _pixels_xy2, 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2 , OPNAME ## _no_rnd_pixels_x2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2 , OPNAME ## _no_rnd_pixels_y2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2, OPNAME ## _no_rnd_pixels_xy2, 8)\
-\
-void (*OPNAME ## _pixels_tab[2][4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- {\
- OPNAME ## _pixels,\
- OPNAME ## _pixels_x2,\
- OPNAME ## _pixels_y2,\
- OPNAME ## _pixels_xy2},\
- {\
- OPNAME ## _pixels16,\
- OPNAME ## _pixels16_x2,\
- OPNAME ## _pixels16_y2,\
- OPNAME ## _pixels16_xy2}\
-};\
-\
-void (*OPNAME ## _no_rnd_pixels_tab[2][4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- {\
- OPNAME ## _pixels,\
- OPNAME ## _no_rnd_pixels_x2,\
- OPNAME ## _no_rnd_pixels_y2,\
- OPNAME ## _no_rnd_pixels_xy2},\
- {\
- OPNAME ## _pixels16,\
- OPNAME ## _no_rnd_pixels16_x2,\
- OPNAME ## _no_rnd_pixels16_y2,\
- OPNAME ## _no_rnd_pixels16_xy2}\
-};
+CALL_2X_PIXELS(OPNAME ## _pixels16_c , OPNAME ## _pixels_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels_x2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels_y2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels_xy2_c, 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels_x2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels_y2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels_xy2_c, 8)
#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEFEFEFEFEULL)>>1) )
#else // 64 bit variant
#define PIXOP2(OPNAME, OP) \
-static void OPNAME ## _pixels8(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+static void OPNAME ## _pixels8_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
int i;\
for(i=0; i<h; i++){\
OP(*((uint32_t*)(block )), LD32(pixels ));\
block +=line_size;\
}\
}\
-static inline void OPNAME ## _no_rnd_pixels8(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
- OPNAME ## _pixels8(block, pixels, line_size, h);\
+static inline void OPNAME ## _no_rnd_pixels8_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels8_c(block, pixels, line_size, h);\
}\
\
static inline void OPNAME ## _no_rnd_pixels8_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
OPNAME ## _no_rnd_pixels8_l2(dst+8, src1+8, src2+8, dst_stride, src_stride1, src_stride2, h);\
}\
\
-static inline void OPNAME ## _no_rnd_pixels8_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+static inline void OPNAME ## _no_rnd_pixels8_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
OPNAME ## _no_rnd_pixels8_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
}\
\
-static inline void OPNAME ## _pixels8_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+static inline void OPNAME ## _pixels8_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
OPNAME ## _pixels8_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
}\
\
-static inline void OPNAME ## _no_rnd_pixels8_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+static inline void OPNAME ## _no_rnd_pixels8_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
OPNAME ## _no_rnd_pixels8_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
-static inline void OPNAME ## _pixels8_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+static inline void OPNAME ## _pixels8_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
OPNAME ## _pixels8_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
OPNAME ## _no_rnd_pixels8_l4(dst+8, src1+8, src2+8, src3+8, src4+8, dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
}\
\
-static inline void OPNAME ## _pixels8_xy2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static inline void OPNAME ## _pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int j;\
for(j=0; j<2; j++){\
}\
}\
\
-static inline void OPNAME ## _no_rnd_pixels8_xy2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static inline void OPNAME ## _no_rnd_pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int j;\
for(j=0; j<2; j++){\
}\
}\
\
-CALL_2X_PIXELS(OPNAME ## _pixels16 , OPNAME ## _pixels8 , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_x2 , OPNAME ## _pixels8_x2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_y2 , OPNAME ## _pixels8_y2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _pixels16_xy2, OPNAME ## _pixels8_xy2, 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16 , OPNAME ## _pixels8 , 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2 , OPNAME ## _no_rnd_pixels8_x2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2 , OPNAME ## _no_rnd_pixels8_y2 , 8)\
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2, OPNAME ## _no_rnd_pixels8_xy2, 8)\
-\
-void (*OPNAME ## _pixels_tab[2][4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- {\
- OPNAME ## _pixels16,\
- OPNAME ## _pixels16_x2,\
- OPNAME ## _pixels16_y2,\
- OPNAME ## _pixels16_xy2},\
- {\
- OPNAME ## _pixels8,\
- OPNAME ## _pixels8_x2,\
- OPNAME ## _pixels8_y2,\
- OPNAME ## _pixels8_xy2},\
-};\
-\
-void (*OPNAME ## _no_rnd_pixels_tab[2][4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- {\
- OPNAME ## _pixels16,\
- OPNAME ## _no_rnd_pixels16_x2,\
- OPNAME ## _no_rnd_pixels16_y2,\
- OPNAME ## _no_rnd_pixels16_xy2},\
- {\
- OPNAME ## _pixels8,\
- OPNAME ## _no_rnd_pixels8_x2,\
- OPNAME ## _no_rnd_pixels8_y2,\
- OPNAME ## _no_rnd_pixels8_xy2},\
-};
+CALL_2X_PIXELS(OPNAME ## _pixels16_c , OPNAME ## _pixels8_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels8_x2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels8_y2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels8_xy2_c, 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_c , OPNAME ## _pixels8_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels8_x2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels8_y2_c , 8)\
+CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels8_xy2_c, 8)\
#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEUL)>>1) )
#endif
#undef op_avg
#undef op_put
-#if 0
-/* FIXME this stuff could be removed as its ot really used anymore */
-#define PIXOP(BTYPE, OPNAME, OP, INCR) \
- \
-static void OPNAME ## _pixels(BTYPE *block, const UINT8 *pixels, int line_size, int h) \
-{ \
- BTYPE *p; \
- const UINT8 *pix; \
- \
- p = block; \
- pix = pixels; \
- do { \
- OP(p[0], pix[0]); \
- OP(p[1], pix[1]); \
- OP(p[2], pix[2]); \
- OP(p[3], pix[3]); \
- OP(p[4], pix[4]); \
- OP(p[5], pix[5]); \
- OP(p[6], pix[6]); \
- OP(p[7], pix[7]); \
- pix += line_size; \
- p += INCR; \
- } while (--h);; \
-} \
- \
-static void OPNAME ## _pixels_x2(BTYPE *block, const UINT8 *pixels, int line_size, int h) \
-{ \
- BTYPE *p; \
- const UINT8 *pix; \
- \
- p = block; \
- pix = pixels; \
- do { \
- OP(p[0], avg2(pix[0], pix[1])); \
- OP(p[1], avg2(pix[1], pix[2])); \
- OP(p[2], avg2(pix[2], pix[3])); \
- OP(p[3], avg2(pix[3], pix[4])); \
- OP(p[4], avg2(pix[4], pix[5])); \
- OP(p[5], avg2(pix[5], pix[6])); \
- OP(p[6], avg2(pix[6], pix[7])); \
- OP(p[7], avg2(pix[7], pix[8])); \
- pix += line_size; \
- p += INCR; \
- } while (--h); \
-} \
- \
-static void OPNAME ## _pixels_y2(BTYPE *block, const UINT8 *pixels, int line_size, int h) \
-{ \
- BTYPE *p; \
- const UINT8 *pix; \
- const UINT8 *pix1; \
- \
- p = block; \
- pix = pixels; \
- pix1 = pixels + line_size; \
- do { \
- OP(p[0], avg2(pix[0], pix1[0])); \
- OP(p[1], avg2(pix[1], pix1[1])); \
- OP(p[2], avg2(pix[2], pix1[2])); \
- OP(p[3], avg2(pix[3], pix1[3])); \
- OP(p[4], avg2(pix[4], pix1[4])); \
- OP(p[5], avg2(pix[5], pix1[5])); \
- OP(p[6], avg2(pix[6], pix1[6])); \
- OP(p[7], avg2(pix[7], pix1[7])); \
- pix += line_size; \
- pix1 += line_size; \
- p += INCR; \
- } while(--h); \
-} \
- \
-static void OPNAME ## _pixels_xy2(BTYPE *block, const UINT8 *pixels, int line_size, int h) \
-{ \
- BTYPE *p; \
- const UINT8 *pix; \
- const UINT8 *pix1; \
- \
- p = block; \
- pix = pixels; \
- pix1 = pixels + line_size; \
- do { \
- OP(p[0], avg4(pix[0], pix[1], pix1[0], pix1[1])); \
- OP(p[1], avg4(pix[1], pix[2], pix1[1], pix1[2])); \
- OP(p[2], avg4(pix[2], pix[3], pix1[2], pix1[3])); \
- OP(p[3], avg4(pix[3], pix[4], pix1[3], pix1[4])); \
- OP(p[4], avg4(pix[4], pix[5], pix1[4], pix1[5])); \
- OP(p[5], avg4(pix[5], pix[6], pix1[5], pix1[6])); \
- OP(p[6], avg4(pix[6], pix[7], pix1[6], pix1[7])); \
- OP(p[7], avg4(pix[7], pix[8], pix1[7], pix1[8])); \
- pix += line_size; \
- pix1 += line_size; \
- p += INCR; \
- } while(--h); \
-} \
- \
-void (*OPNAME ## _pixels_tab[4])(BTYPE *block, const UINT8 *pixels, int line_size, int h) = { \
- OPNAME ## _pixels, \
- OPNAME ## _pixels_x2, \
- OPNAME ## _pixels_y2, \
- OPNAME ## _pixels_xy2, \
-};
-
-/* rounding primitives */
#define avg2(a,b) ((a+b+1)>>1)
#define avg4(a,b,c,d) ((a+b+c+d+2)>>2)
-#define op_avg(a, b) a = avg2(a, b)
-#define op_sub(a, b) a -= b
-#define op_put(a, b) a = b
-
-PIXOP(DCTELEM, sub, op_sub, 8)
-PIXOP(uint8_t, avg, op_avg, line_size)
-PIXOP(uint8_t, put, op_put, line_size)
-
-/* not rounding primitives */
-#undef avg2
-#undef avg4
-#define avg2(a,b) ((a+b)>>1)
-#define avg4(a,b,c,d) ((a+b+c+d+1)>>2)
-
-PIXOP(uint8_t, avg_no_rnd, op_avg, line_size)
-PIXOP(uint8_t, put_no_rnd, op_put, line_size)
-/* motion estimation */
-
-#undef avg2
-#undef avg4
-#endif
-
-#define avg2(a,b) ((a+b+1)>>1)
-#define avg4(a,b,c,d) ((a+b+c+d+2)>>2)
static void gmc1_c(UINT8 *dst, UINT8 *src, int stride, int h, int x16, int y16, int rounder)
{
const int C=(16-x16)*( y16);
const int D=( x16)*( y16);
int i;
- rounder= 128 - rounder;
for(i=0; i<h; i++)
{
}
}
+static void gmc_c(UINT8 *dst, UINT8 *src, int stride, int h, int ox, int oy,
+ int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
+{
+ int y, vx, vy;
+ const int s= 1<<shift;
+
+ width--;
+ height--;
+
+ for(y=0; y<h; y++){
+ int x;
+
+ vx= ox;
+ vy= oy;
+ for(x=0; x<8; x++){ //XXX FIXME optimize
+ int src_x, src_y, frac_x, frac_y, index;
+
+ src_x= vx>>16;
+ src_y= vy>>16;
+ frac_x= src_x&(s-1);
+ frac_y= src_y&(s-1);
+ src_x>>=shift;
+ src_y>>=shift;
+
+ if((unsigned)src_x < width){
+ if((unsigned)src_y < height){
+ index= src_x + src_y*stride;
+ dst[y*stride + x]= ( ( src[index ]*(s-frac_x)
+ + src[index +1]* frac_x )*(s-frac_y)
+ + ( src[index+stride ]*(s-frac_x)
+ + src[index+stride+1]* frac_x )* frac_y
+ + r)>>(shift*2);
+ }else{
+ index= src_x + clip(src_y, 0, height)*stride;
+ dst[y*stride + x]= ( ( src[index ]*(s-frac_x)
+ + src[index +1]* frac_x )*s
+ + r)>>(shift*2);
+ }
+ }else{
+ if((unsigned)src_y < height){
+ index= clip(src_x, 0, width) + src_y*stride;
+ dst[y*stride + x]= ( ( src[index ]*(s-frac_y)
+ + src[index+stride ]* frac_y )*s
+ + r)>>(shift*2);
+ }else{
+ index= clip(src_x, 0, width) + clip(src_y, 0, height)*stride;
+ dst[y*stride + x]= src[index ];
+ }
+ }
+
+ vx+= dxx;
+ vy+= dyx;
+ }
+ ox += dxy;
+ oy += dyy;
+ }
+}
+
static inline void copy_block17(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h)
{
int i;
}\
\
static void OPNAME ## qpel8_mc00_c (UINT8 *dst, UINT8 *src, int stride){\
- OPNAME ## pixels8(dst, src, stride, 8);\
+ OPNAME ## pixels8_c(dst, src, stride, 8);\
}\
\
static void OPNAME ## qpel8_mc10_c(UINT8 *dst, UINT8 *src, int stride){\
OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8, 8);\
}\
static void OPNAME ## qpel16_mc00_c (UINT8 *dst, UINT8 *src, int stride){\
- OPNAME ## pixels16(dst, src, stride, 16);\
+ OPNAME ## pixels16_c(dst, src, stride, 16);\
}\
\
static void OPNAME ## qpel16_mc10_c(UINT8 *dst, UINT8 *src, int stride){\
UINT8 halfH[272];\
put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16, 16);\
-}\
-qpel_mc_func OPNAME ## qpel_pixels_tab[2][16]={ \
- {\
- OPNAME ## qpel16_mc00_c, \
- OPNAME ## qpel16_mc10_c, \
- OPNAME ## qpel16_mc20_c, \
- OPNAME ## qpel16_mc30_c, \
- OPNAME ## qpel16_mc01_c, \
- OPNAME ## qpel16_mc11_c, \
- OPNAME ## qpel16_mc21_c, \
- OPNAME ## qpel16_mc31_c, \
- OPNAME ## qpel16_mc02_c, \
- OPNAME ## qpel16_mc12_c, \
- OPNAME ## qpel16_mc22_c, \
- OPNAME ## qpel16_mc32_c, \
- OPNAME ## qpel16_mc03_c, \
- OPNAME ## qpel16_mc13_c, \
- OPNAME ## qpel16_mc23_c, \
- OPNAME ## qpel16_mc33_c, \
- },{\
- OPNAME ## qpel8_mc00_c, \
- OPNAME ## qpel8_mc10_c, \
- OPNAME ## qpel8_mc20_c, \
- OPNAME ## qpel8_mc30_c, \
- OPNAME ## qpel8_mc01_c, \
- OPNAME ## qpel8_mc11_c, \
- OPNAME ## qpel8_mc21_c, \
- OPNAME ## qpel8_mc31_c, \
- OPNAME ## qpel8_mc02_c, \
- OPNAME ## qpel8_mc12_c, \
- OPNAME ## qpel8_mc22_c, \
- OPNAME ## qpel8_mc32_c, \
- OPNAME ## qpel8_mc03_c, \
- OPNAME ## qpel8_mc13_c, \
- OPNAME ## qpel8_mc23_c, \
- OPNAME ## qpel8_mc33_c, \
- }\
-};
+}
#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1)
#define op_avg_no_rnd(a, b) a = (((a)+cm[((b) + 15)>>5])>>1)
#undef op_put
#undef op_put_no_rnd
-int pix_abs16x16_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+ int i;
+
+ for(i=0; i<h; i++){
+ dst[0]= cm[(9*(src[0] + src[1]) - (src[-1] + src[2]) + 8)>>4];
+ dst[1]= cm[(9*(src[1] + src[2]) - (src[ 0] + src[3]) + 8)>>4];
+ dst[2]= cm[(9*(src[2] + src[3]) - (src[ 1] + src[4]) + 8)>>4];
+ dst[3]= cm[(9*(src[3] + src[4]) - (src[ 2] + src[5]) + 8)>>4];
+ dst[4]= cm[(9*(src[4] + src[5]) - (src[ 3] + src[6]) + 8)>>4];
+ dst[5]= cm[(9*(src[5] + src[6]) - (src[ 4] + src[7]) + 8)>>4];
+ dst[6]= cm[(9*(src[6] + src[7]) - (src[ 5] + src[8]) + 8)>>4];
+ dst[7]= cm[(9*(src[7] + src[8]) - (src[ 6] + src[9]) + 8)>>4];
+ dst+=dstStride;
+ src+=srcStride;
+ }
+}
+
+static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w){
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+ int i;
+
+ for(i=0; i<w; i++){
+ const int src_1= src[ -srcStride];
+ const int src0 = src[0 ];
+ const int src1 = src[ srcStride];
+ const int src2 = src[2*srcStride];
+ const int src3 = src[3*srcStride];
+ const int src4 = src[4*srcStride];
+ const int src5 = src[5*srcStride];
+ const int src6 = src[6*srcStride];
+ const int src7 = src[7*srcStride];
+ const int src8 = src[8*srcStride];
+ const int src9 = src[9*srcStride];
+ dst[0*dstStride]= cm[(9*(src0 + src1) - (src_1 + src2) + 8)>>4];
+ dst[1*dstStride]= cm[(9*(src1 + src2) - (src0 + src3) + 8)>>4];
+ dst[2*dstStride]= cm[(9*(src2 + src3) - (src1 + src4) + 8)>>4];
+ dst[3*dstStride]= cm[(9*(src3 + src4) - (src2 + src5) + 8)>>4];
+ dst[4*dstStride]= cm[(9*(src4 + src5) - (src3 + src6) + 8)>>4];
+ dst[5*dstStride]= cm[(9*(src5 + src6) - (src4 + src7) + 8)>>4];
+ dst[6*dstStride]= cm[(9*(src6 + src7) - (src5 + src8) + 8)>>4];
+ dst[7*dstStride]= cm[(9*(src7 + src8) - (src6 + src9) + 8)>>4];
+ src++;
+ dst++;
+ }
+}
+
+static void put_mspel8_mc00_c (uint8_t *dst, uint8_t *src, int stride){
+ put_pixels8_c(dst, src, stride, 8);
+}
+
+static void put_mspel8_mc10_c(uint8_t *dst, uint8_t *src, int stride){
+ uint8_t half[64];
+ wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
+ put_pixels8_l2(dst, src, half, stride, stride, 8, 8);
+}
+
+static void put_mspel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){
+ wmv2_mspel8_h_lowpass(dst, src, stride, stride, 8);
+}
+
+static void put_mspel8_mc30_c(uint8_t *dst, uint8_t *src, int stride){
+ uint8_t half[64];
+ wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
+ put_pixels8_l2(dst, src+1, half, stride, stride, 8, 8);
+}
+
+static void put_mspel8_mc02_c(uint8_t *dst, uint8_t *src, int stride){
+ wmv2_mspel8_v_lowpass(dst, src, stride, stride, 8);
+}
+
+static void put_mspel8_mc12_c(uint8_t *dst, uint8_t *src, int stride){
+ uint8_t halfH[88];
+ uint8_t halfV[64];
+ uint8_t halfHV[64];
+ wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
+ wmv2_mspel8_v_lowpass(halfV, src, 8, stride, 8);
+ wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8);
+ put_pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);
+}
+static void put_mspel8_mc32_c(uint8_t *dst, uint8_t *src, int stride){
+ uint8_t halfH[88];
+ uint8_t halfV[64];
+ uint8_t halfHV[64];
+ wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
+ wmv2_mspel8_v_lowpass(halfV, src+1, 8, stride, 8);
+ wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8);
+ put_pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);
+}
+static void put_mspel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){
+ uint8_t halfH[88];
+ wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
+ wmv2_mspel8_v_lowpass(dst, halfH+8, stride, 8, 8);
+}
+
+
+static inline int pix_abs16x16_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
return s;
}
-int pix_abs16x16_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
return s;
}
-int pix_abs16x16_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
UINT8 *pix3 = pix2 + line_size;
return s;
}
-int pix_abs16x16_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
UINT8 *pix3 = pix2 + line_size;
return s;
}
-int pix_abs8x8_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static inline int pix_abs8x8_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
return s;
}
-int pix_abs8x8_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
return s;
}
-int pix_abs8x8_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
UINT8 *pix3 = pix2 + line_size;
return s;
}
-int pix_abs8x8_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
{
int s, i;
UINT8 *pix3 = pix2 + line_size;
return s;
}
-/* permute block according so that it corresponds to the MMX idct
- order */
-void block_permute(INT16 *block, UINT8 *permutation)
-{
- int i;
- INT16 temp[64];
+static int sad16x16_c(void *s, uint8_t *a, uint8_t *b, int stride){
+ return pix_abs16x16_c(a,b,stride);
+}
- for(i=0; i<64; i++) temp[ permutation[i] ] = block[i];
+static int sad8x8_c(void *s, uint8_t *a, uint8_t *b, int stride){
+ return pix_abs8x8_c(a,b,stride);
+}
- for(i=0; i<64; i++) block[i] = temp[i];
+void ff_block_permute(DCTELEM *block, UINT8 *permutation, const UINT8 *scantable, int last)
+{
+ int i;
+ DCTELEM temp[64];
+
+ if(last<=0) return;
+ //if(permutation[1]==1) return; //FIXME its ok but not clean and might fail for some perms
+
+ for(i=0; i<=last; i++){
+ const int j= scantable[i];
+ temp[j]= block[j];
+ block[j]=0;
+ }
+
+ for(i=0; i<=last; i++){
+ const int j= scantable[i];
+ const int perm_j= permutation[j];
+ block[perm_j]= temp[j];
+ }
}
-void clear_blocks_c(DCTELEM *blocks)
+static void clear_blocks_c(DCTELEM *blocks)
{
memset(blocks, 0, sizeof(DCTELEM)*6*64);
}
-void dsputil_init(void)
-{
- int i, j;
+static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){
+ int i;
+ for(i=0; i+7<w; i++){
+ dst[i+0] += src[i+0];
+ dst[i+1] += src[i+1];
+ dst[i+2] += src[i+2];
+ dst[i+3] += src[i+3];
+ dst[i+4] += src[i+4];
+ dst[i+5] += src[i+5];
+ dst[i+6] += src[i+6];
+ dst[i+7] += src[i+7];
+ }
+ for(; i<w; i++)
+ dst[i+0] += src[i+0];
+}
- for(i=0;i<256;i++) cropTbl[i + MAX_NEG_CROP] = i;
- for(i=0;i<MAX_NEG_CROP;i++) {
- cropTbl[i] = 0;
- cropTbl[i + MAX_NEG_CROP + 256] = 255;
+static void diff_bytes_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
+ int i;
+ for(i=0; i+7<w; i++){
+ dst[i+0] = src1[i+0]-src2[i+0];
+ dst[i+1] = src1[i+1]-src2[i+1];
+ dst[i+2] = src1[i+2]-src2[i+2];
+ dst[i+3] = src1[i+3]-src2[i+3];
+ dst[i+4] = src1[i+4]-src2[i+4];
+ dst[i+5] = src1[i+5]-src2[i+5];
+ dst[i+6] = src1[i+6]-src2[i+6];
+ dst[i+7] = src1[i+7]-src2[i+7];
}
+ for(; i<w; i++)
+ dst[i+0] = src1[i+0]-src2[i+0];
+}
- for(i=0;i<512;i++) {
- squareTbl[i] = (i - 256) * (i - 256);
+#define BUTTERFLY2(o1,o2,i1,i2) \
+o1= (i1)+(i2);\
+o2= (i1)-(i2);
+
+#define BUTTERFLY1(x,y) \
+{\
+ int a,b;\
+ a= x;\
+ b= y;\
+ x= a+b;\
+ y= a-b;\
+}
+
+#define BUTTERFLYA(x,y) (ABS((x)+(y)) + ABS((x)-(y)))
+
+static int hadamard8_diff_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride){
+ int i;
+ int temp[64];
+ int sum=0;
+
+ for(i=0; i<8; i++){
+ //FIXME try pointer walks
+ BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0]-dst[stride*i+0],src[stride*i+1]-dst[stride*i+1]);
+ BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2]-dst[stride*i+2],src[stride*i+3]-dst[stride*i+3]);
+ BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4]-dst[stride*i+4],src[stride*i+5]-dst[stride*i+5]);
+ BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6]-dst[stride*i+6],src[stride*i+7]-dst[stride*i+7]);
+
+ BUTTERFLY1(temp[8*i+0], temp[8*i+2]);
+ BUTTERFLY1(temp[8*i+1], temp[8*i+3]);
+ BUTTERFLY1(temp[8*i+4], temp[8*i+6]);
+ BUTTERFLY1(temp[8*i+5], temp[8*i+7]);
+
+ BUTTERFLY1(temp[8*i+0], temp[8*i+4]);
+ BUTTERFLY1(temp[8*i+1], temp[8*i+5]);
+ BUTTERFLY1(temp[8*i+2], temp[8*i+6]);
+ BUTTERFLY1(temp[8*i+3], temp[8*i+7]);
+ }
+
+ for(i=0; i<8; i++){
+ BUTTERFLY1(temp[8*0+i], temp[8*1+i]);
+ BUTTERFLY1(temp[8*2+i], temp[8*3+i]);
+ BUTTERFLY1(temp[8*4+i], temp[8*5+i]);
+ BUTTERFLY1(temp[8*6+i], temp[8*7+i]);
+
+ BUTTERFLY1(temp[8*0+i], temp[8*2+i]);
+ BUTTERFLY1(temp[8*1+i], temp[8*3+i]);
+ BUTTERFLY1(temp[8*4+i], temp[8*6+i]);
+ BUTTERFLY1(temp[8*5+i], temp[8*7+i]);
+
+ sum +=
+ BUTTERFLYA(temp[8*0+i], temp[8*4+i])
+ +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
+ +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
+ +BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
+ }
+#if 0
+static int maxi=0;
+if(sum>maxi){
+ maxi=sum;
+ printf("MAX:%d\n", maxi);
+}
+#endif
+ return sum;
+}
+
+static int hadamard8_abs_c(uint8_t *src, int stride, int mean){
+ int i;
+ int temp[64];
+ int sum=0;
+//FIXME OOOPS ignore 0 term instead of mean mess
+ for(i=0; i<8; i++){
+ //FIXME try pointer walks
+ BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0]-mean,src[stride*i+1]-mean);
+ BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2]-mean,src[stride*i+3]-mean);
+ BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4]-mean,src[stride*i+5]-mean);
+ BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6]-mean,src[stride*i+7]-mean);
+
+ BUTTERFLY1(temp[8*i+0], temp[8*i+2]);
+ BUTTERFLY1(temp[8*i+1], temp[8*i+3]);
+ BUTTERFLY1(temp[8*i+4], temp[8*i+6]);
+ BUTTERFLY1(temp[8*i+5], temp[8*i+7]);
+
+ BUTTERFLY1(temp[8*i+0], temp[8*i+4]);
+ BUTTERFLY1(temp[8*i+1], temp[8*i+5]);
+ BUTTERFLY1(temp[8*i+2], temp[8*i+6]);
+ BUTTERFLY1(temp[8*i+3], temp[8*i+7]);
+ }
+
+ for(i=0; i<8; i++){
+ BUTTERFLY1(temp[8*0+i], temp[8*1+i]);
+ BUTTERFLY1(temp[8*2+i], temp[8*3+i]);
+ BUTTERFLY1(temp[8*4+i], temp[8*5+i]);
+ BUTTERFLY1(temp[8*6+i], temp[8*7+i]);
+
+ BUTTERFLY1(temp[8*0+i], temp[8*2+i]);
+ BUTTERFLY1(temp[8*1+i], temp[8*3+i]);
+ BUTTERFLY1(temp[8*4+i], temp[8*6+i]);
+ BUTTERFLY1(temp[8*5+i], temp[8*7+i]);
+
+ sum +=
+ BUTTERFLYA(temp[8*0+i], temp[8*4+i])
+ +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
+ +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
+ +BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
}
+
+ return sum;
+}
- get_pixels = get_pixels_c;
- diff_pixels = diff_pixels_c;
- put_pixels_clamped = put_pixels_clamped_c;
- add_pixels_clamped = add_pixels_clamped_c;
- gmc1= gmc1_c;
- clear_blocks= clear_blocks_c;
- pix_sum= pix_sum_c;
- pix_norm1= pix_norm1_c;
-
- pix_abs16x16 = pix_abs16x16_c;
- pix_abs16x16_x2 = pix_abs16x16_x2_c;
- pix_abs16x16_y2 = pix_abs16x16_y2_c;
- pix_abs16x16_xy2 = pix_abs16x16_xy2_c;
- pix_abs8x8 = pix_abs8x8_c;
- pix_abs8x8_x2 = pix_abs8x8_x2_c;
- pix_abs8x8_y2 = pix_abs8x8_y2_c;
- pix_abs8x8_xy2 = pix_abs8x8_xy2_c;
+static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ DCTELEM temp[64];
+ int sum=0, i;
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+ s->fdct(temp);
+
+ for(i=0; i<64; i++)
+ sum+= ABS(temp[i]);
+
+ return sum;
+}
+
+void simple_idct(INT16 *block); //FIXME
+
+static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ DCTELEM temp[64], bak[64];
+ int sum=0, i;
+
+ s->mb_intra=0;
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+
+ memcpy(bak, temp, 64*sizeof(DCTELEM));
+
+ s->dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
+ s->dct_unquantize(s, temp, 0, s->qscale);
+ simple_idct(temp); //FIXME
+
+ for(i=0; i<64; i++)
+ sum+= (temp[i]-bak[i])*(temp[i]-bak[i]);
+
+ return sum;
+}
+
+WARPER88_1616(hadamard8_diff_c, hadamard8_diff16_c)
+WARPER88_1616(dct_sad8x8_c, dct_sad16x16_c)
+WARPER88_1616(quant_psnr8x8_c, quant_psnr16x16_c)
+
+void dsputil_init(DSPContext* c, unsigned mask)
+{
+ static int init_done = 0;
+ int i;
+
+ if (!init_done) {
+ for(i=0;i<256;i++) cropTbl[i + MAX_NEG_CROP] = i;
+ for(i=0;i<MAX_NEG_CROP;i++) {
+ cropTbl[i] = 0;
+ cropTbl[i + MAX_NEG_CROP + 256] = 255;
+ }
+
+ for(i=0;i<512;i++) {
+ squareTbl[i] = (i - 256) * (i - 256);
+ }
+
+ for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1;
+
+ init_done = 1;
+ }
+
+ c->get_pixels = get_pixels_c;
+ c->diff_pixels = diff_pixels_c;
+ c->put_pixels_clamped = put_pixels_clamped_c;
+ c->add_pixels_clamped = add_pixels_clamped_c;
+ c->gmc1 = gmc1_c;
+ c->gmc = gmc_c;
+ c->clear_blocks = clear_blocks_c;
+ c->pix_sum = pix_sum_c;
+ c->pix_norm1 = pix_norm1_c;
+ c->sse[0]= sse16_c;
+ c->sse[1]= sse8_c;
+
+ /* TODO [0] 16 [1] 8 */
+ c->pix_abs16x16 = pix_abs16x16_c;
+ c->pix_abs16x16_x2 = pix_abs16x16_x2_c;
+ c->pix_abs16x16_y2 = pix_abs16x16_y2_c;
+ c->pix_abs16x16_xy2 = pix_abs16x16_xy2_c;
+ c->pix_abs8x8 = pix_abs8x8_c;
+ c->pix_abs8x8_x2 = pix_abs8x8_x2_c;
+ c->pix_abs8x8_y2 = pix_abs8x8_y2_c;
+ c->pix_abs8x8_xy2 = pix_abs8x8_xy2_c;
+
+#define dspfunc(PFX, IDX, NUM) \
+ c->PFX ## _pixels_tab[IDX][0] = PFX ## _pixels ## NUM ## _c; \
+ c->PFX ## _pixels_tab[IDX][1] = PFX ## _pixels ## NUM ## _x2_c; \
+ c->PFX ## _pixels_tab[IDX][2] = PFX ## _pixels ## NUM ## _y2_c; \
+ c->PFX ## _pixels_tab[IDX][3] = PFX ## _pixels ## NUM ## _xy2_c
+
+ dspfunc(put, 0, 16);
+ dspfunc(put_no_rnd, 0, 16);
+ dspfunc(put, 1, 8);
+ dspfunc(put_no_rnd, 1, 8);
+
+ dspfunc(avg, 0, 16);
+ dspfunc(avg_no_rnd, 0, 16);
+ dspfunc(avg, 1, 8);
+ dspfunc(avg_no_rnd, 1, 8);
+#undef dspfunc
+
+#define dspfunc(PFX, IDX, NUM) \
+ c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_c; \
+ c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_c; \
+ c->PFX ## _pixels_tab[IDX][ 2] = PFX ## NUM ## _mc20_c; \
+ c->PFX ## _pixels_tab[IDX][ 3] = PFX ## NUM ## _mc30_c; \
+ c->PFX ## _pixels_tab[IDX][ 4] = PFX ## NUM ## _mc01_c; \
+ c->PFX ## _pixels_tab[IDX][ 5] = PFX ## NUM ## _mc11_c; \
+ c->PFX ## _pixels_tab[IDX][ 6] = PFX ## NUM ## _mc21_c; \
+ c->PFX ## _pixels_tab[IDX][ 7] = PFX ## NUM ## _mc31_c; \
+ c->PFX ## _pixels_tab[IDX][ 8] = PFX ## NUM ## _mc02_c; \
+ c->PFX ## _pixels_tab[IDX][ 9] = PFX ## NUM ## _mc12_c; \
+ c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_c; \
+ c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_c; \
+ c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_c; \
+ c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_c; \
+ c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_c; \
+ c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_c
+
+ dspfunc(put_qpel, 0, 16);
+ dspfunc(put_no_rnd_qpel, 0, 16);
+
+ dspfunc(avg_qpel, 0, 16);
+ /* dspfunc(avg_no_rnd_qpel, 0, 16); */
+
+ dspfunc(put_qpel, 1, 8);
+ dspfunc(put_no_rnd_qpel, 1, 8);
+
+ dspfunc(avg_qpel, 1, 8);
+ /* dspfunc(avg_no_rnd_qpel, 1, 8); */
+#undef dspfunc
+
+ c->put_mspel_pixels_tab[0]= put_mspel8_mc00_c;
+ c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
+ c->put_mspel_pixels_tab[2]= put_mspel8_mc20_c;
+ c->put_mspel_pixels_tab[3]= put_mspel8_mc30_c;
+ c->put_mspel_pixels_tab[4]= put_mspel8_mc02_c;
+ c->put_mspel_pixels_tab[5]= put_mspel8_mc12_c;
+ c->put_mspel_pixels_tab[6]= put_mspel8_mc22_c;
+ c->put_mspel_pixels_tab[7]= put_mspel8_mc32_c;
+
+ c->hadamard8_diff[0]= hadamard8_diff16_c;
+ c->hadamard8_diff[1]= hadamard8_diff_c;
+ c->hadamard8_abs = hadamard8_abs_c;
+
+ c->dct_sad[0]= dct_sad16x16_c;
+ c->dct_sad[1]= dct_sad8x8_c;
+
+ c->sad[0]= sad16x16_c;
+ c->sad[1]= sad8x8_c;
+
+ c->quant_psnr[0]= quant_psnr16x16_c;
+ c->quant_psnr[1]= quant_psnr8x8_c;
+
+ c->add_bytes= add_bytes_c;
+ c->diff_bytes= diff_bytes_c;
#ifdef HAVE_MMX
- dsputil_init_mmx();
+ dsputil_init_mmx(c, mask);
+ if (ff_bit_exact)
+ {
+ /* FIXME - AVCodec context should have flag for bitexact match */
+ /* fprintf(stderr, "\n\n\nff_bit_exact %d\n\n\n\n", ff_bit_exact); */
+ dsputil_set_bit_exact_mmx(c, mask);
+ }
#endif
#ifdef ARCH_ARMV4L
- dsputil_init_armv4l();
+ dsputil_init_armv4l(c, mask);
#endif
#ifdef HAVE_MLIB
- dsputil_init_mlib();
+ dsputil_init_mlib(c, mask);
#endif
#ifdef ARCH_ALPHA
- dsputil_init_alpha();
+ dsputil_init_alpha(c, mask);
#endif
#ifdef ARCH_POWERPC
- dsputil_init_ppc();
+ dsputil_init_ppc(c, mask);
#endif
#ifdef HAVE_MMI
- dsputil_init_mmi();
+ dsputil_init_mmi(c, mask);
#endif
-
- for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1;
}
/* remove any non bit exact operation (testing purpose) */
{
ff_bit_exact=1;
#ifdef HAVE_MMX
- dsputil_set_bit_exact_mmx();
+// FIXME - better set_bit_exact
+// dsputil_set_bit_exact_mmx();
#endif
}
-
-void get_psnr(UINT8 *orig_image[3], UINT8 *coded_image[3],
- int orig_linesize[3], int coded_linesize,
- AVCodecContext *avctx)
-{
- int quad, diff, x, y;
- UINT8 *orig, *coded;
- UINT32 *sq = squareTbl + 256;
-
- quad = 0;
- diff = 0;
-
- /* Luminance */
- orig = orig_image[0];
- coded = coded_image[0];
-
- for (y=0;y<avctx->height;y++) {
- for (x=0;x<avctx->width;x++) {
- diff = *(orig + x) - *(coded + x);
- quad += sq[diff];
- }
- orig += orig_linesize[0];
- coded += coded_linesize;
- }
-
- avctx->psnr_y = (float) quad / (float) (avctx->width * avctx->height);
-
- if (avctx->psnr_y) {
- avctx->psnr_y = (float) (255 * 255) / avctx->psnr_y;
- avctx->psnr_y = 10 * (float) log10 (avctx->psnr_y);
- } else
- avctx->psnr_y = 99.99;
-}
-