/*
* DSP utils
* Copyright (c) 2000, 2001 Fabrice Bellard.
+ * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
*
* gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
*/
+
+/**
+ * @file dsputil.c
+ * DSP utils
+ */
+
#include "avcodec.h"
#include "dsputil.h"
+#include "mpegvideo.h"
#include "simple_idct.h"
+#include "faandct.h"
+
+/* snow.c */
+void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type, int decomposition_count);
+
+uint8_t cropTbl[256 + 2 * MAX_NEG_CROP] = {0, };
+uint32_t squareTbl[512] = {0, };
-void (*ff_idct)(DCTELEM *block);
-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);
-
-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;
-
-UINT8 cropTbl[256 + 2 * MAX_NEG_CROP];
-UINT32 squareTbl[512];
-
-extern INT16 default_intra_matrix[64];
-extern INT16 default_non_intra_matrix[64];
-extern INT16 ff_mpeg4_default_intra_matrix[64];
-extern INT16 ff_mpeg4_default_non_intra_matrix[64];
-
-UINT8 zigzag_direct[64] = {
- 0, 1, 8, 16, 9, 2, 3, 10,
- 17, 24, 32, 25, 18, 11, 4, 5,
+const uint8_t ff_zigzag_direct[64] = {
+ 0, 1, 8, 16, 9, 2, 3, 10,
+ 17, 24, 32, 25, 18, 11, 4, 5,
12, 19, 26, 33, 40, 48, 41, 34,
- 27, 20, 13, 6, 7, 14, 21, 28,
+ 27, 20, 13, 6, 7, 14, 21, 28,
35, 42, 49, 56, 57, 50, 43, 36,
29, 22, 15, 23, 30, 37, 44, 51,
58, 59, 52, 45, 38, 31, 39, 46,
53, 60, 61, 54, 47, 55, 62, 63
};
-/* not permutated inverse zigzag_direct + 1 for MMX quantizer */
-UINT16 __align8 inv_zigzag_direct16[64];
+/* Specific zigzag scan for 248 idct. NOTE that unlike the
+ specification, we interleave the fields */
+const uint8_t ff_zigzag248_direct[64] = {
+ 0, 8, 1, 9, 16, 24, 2, 10,
+ 17, 25, 32, 40, 48, 56, 33, 41,
+ 18, 26, 3, 11, 4, 12, 19, 27,
+ 34, 42, 49, 57, 50, 58, 35, 43,
+ 20, 28, 5, 13, 6, 14, 21, 29,
+ 36, 44, 51, 59, 52, 60, 37, 45,
+ 22, 30, 7, 15, 23, 31, 38, 46,
+ 53, 61, 54, 62, 39, 47, 55, 63,
+};
-/* not permutated zigzag_direct for MMX quantizer */
-UINT8 zigzag_direct_noperm[64];
+/* not permutated inverse zigzag_direct + 1 for MMX quantizer */
+uint16_t __align8 inv_zigzag_direct16[64] = {0, };
-UINT8 ff_alternate_horizontal_scan[64] = {
- 0, 1, 2, 3, 8, 9, 16, 17,
+const uint8_t ff_alternate_horizontal_scan[64] = {
+ 0, 1, 2, 3, 8, 9, 16, 17,
10, 11, 4, 5, 6, 7, 15, 14,
13, 12, 19, 18, 24, 25, 32, 33,
26, 27, 20, 21, 22, 23, 28, 29,
52, 53, 54, 55, 60, 61, 62, 63,
};
-UINT8 ff_alternate_vertical_scan[64] = {
- 0, 8, 16, 24, 1, 9, 2, 10,
+const uint8_t ff_alternate_vertical_scan[64] = {
+ 0, 8, 16, 24, 1, 9, 2, 10,
17, 25, 32, 40, 48, 56, 57, 49,
41, 33, 26, 18, 3, 11, 4, 12,
19, 27, 34, 42, 50, 58, 35, 43,
38, 46, 54, 62, 39, 47, 55, 63,
};
-#ifdef SIMPLE_IDCT
-
-/* Input permutation for the simple_idct_mmx */
-static UINT8 simple_mmx_permutation[64]={
- 0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D,
- 0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D,
- 0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D,
- 0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F,
- 0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F,
- 0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D,
- 0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F,
- 0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
-};
-#endif
-
/* a*inverse[b]>>32 == a/b for all 0<=a<=65536 && 2<=b<=255 */
-UINT32 inverse[256]={
+const uint32_t 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,
};
-/* used to skip zeros at the end */
-UINT8 zigzag_end[64];
+/* Input permutation for the simple_idct_mmx */
+static const uint8_t simple_mmx_permutation[64]={
+ 0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D,
+ 0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D,
+ 0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D,
+ 0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F,
+ 0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F,
+ 0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D,
+ 0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F,
+ 0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
+};
+
+static int pix_sum_c(uint8_t * pix, int line_size)
+{
+ int s, i, j;
+
+ s = 0;
+ for (i = 0; i < 16; i++) {
+ for (j = 0; j < 16; j += 8) {
+ s += pix[0];
+ s += pix[1];
+ s += pix[2];
+ s += pix[3];
+ s += pix[4];
+ s += pix[5];
+ s += pix[6];
+ s += pix[7];
+ pix += 8;
+ }
+ pix += line_size - 16;
+ }
+ return s;
+}
+
+static int pix_norm1_c(uint8_t * pix, int line_size)
+{
+ int s, i, j;
+ uint32_t *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < 16; i++) {
+ for (j = 0; j < 16; j += 8) {
+#if 0
+ s += sq[pix[0]];
+ s += sq[pix[1]];
+ s += sq[pix[2]];
+ s += sq[pix[3]];
+ s += sq[pix[4]];
+ s += sq[pix[5]];
+ s += sq[pix[6]];
+ s += sq[pix[7]];
+#else
+#if LONG_MAX > 2147483647
+ register uint64_t x=*(uint64_t*)pix;
+ s += sq[x&0xff];
+ s += sq[(x>>8)&0xff];
+ s += sq[(x>>16)&0xff];
+ s += sq[(x>>24)&0xff];
+ s += sq[(x>>32)&0xff];
+ s += sq[(x>>40)&0xff];
+ s += sq[(x>>48)&0xff];
+ s += sq[(x>>56)&0xff];
+#else
+ register uint32_t x=*(uint32_t*)pix;
+ s += sq[x&0xff];
+ s += sq[(x>>8)&0xff];
+ s += sq[(x>>16)&0xff];
+ s += sq[(x>>24)&0xff];
+ x=*(uint32_t*)(pix+4);
+ s += sq[x&0xff];
+ s += sq[(x>>8)&0xff];
+ s += sq[(x>>16)&0xff];
+ s += sq[(x>>24)&0xff];
+#endif
+#endif
+ pix += 8;
+ }
+ pix += line_size - 16;
+ }
+ return s;
+}
-UINT8 permutation[64];
-//UINT8 invPermutation[64];
+static void bswap_buf(uint32_t *dst, uint32_t *src, int w){
+ int i;
+
+ for(i=0; i+8<=w; i+=8){
+ dst[i+0]= bswap_32(src[i+0]);
+ dst[i+1]= bswap_32(src[i+1]);
+ dst[i+2]= bswap_32(src[i+2]);
+ dst[i+3]= bswap_32(src[i+3]);
+ dst[i+4]= bswap_32(src[i+4]);
+ dst[i+5]= bswap_32(src[i+5]);
+ dst[i+6]= bswap_32(src[i+6]);
+ dst[i+7]= bswap_32(src[i+7]);
+ }
+ for(;i<w; i++){
+ dst[i+0]= bswap_32(src[i+0]);
+ }
+}
-static void build_zigzag_end(void)
+static int sse4_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
{
- int lastIndex;
- int lastIndexAfterPerm=0;
- for(lastIndex=0; lastIndex<64; lastIndex++)
- {
- if(zigzag_direct[lastIndex] > lastIndexAfterPerm)
- lastIndexAfterPerm= zigzag_direct[lastIndex];
- zigzag_end[lastIndex]= lastIndexAfterPerm + 1;
+ int s, i;
+ uint32_t *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < h; 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]];
+ pix1 += line_size;
+ pix2 += line_size;
+ }
+ return s;
+}
+
+static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
+{
+ int s, i;
+ uint32_t *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < h; 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_t *pix1, uint8_t *pix2, int line_size, int h)
+{
+ int s, i;
+ uint32_t *sq = squareTbl + 256;
+
+ s = 0;
+ for (i = 0; i < h; 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]];
+ s += sq[pix1[ 8] - pix2[ 8]];
+ s += sq[pix1[ 9] - pix2[ 9]];
+ s += sq[pix1[10] - pix2[10]];
+ s += sq[pix1[11] - pix2[11]];
+ s += sq[pix1[12] - pix2[12]];
+ s += sq[pix1[13] - pix2[13]];
+ s += sq[pix1[14] - pix2[14]];
+ s += sq[pix1[15] - pix2[15]];
+
+ pix1 += line_size;
+ pix2 += line_size;
+ }
+ return s;
+}
+
+
+static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int w, int h, int type){
+ int s, i, j;
+ const int dec_count= w==8 ? 3 : 4;
+ int tmp[16*16];
+#if 0
+ int level, ori;
+ static const int scale[2][2][4][4]={
+ {
+ {
+ //8x8 dec=3
+ {268, 239, 239, 213},
+ { 0, 224, 224, 152},
+ { 0, 135, 135, 110},
+ },{
+ //16x16 dec=4
+ {344, 310, 310, 280},
+ { 0, 320, 320, 228},
+ { 0, 175, 175, 136},
+ { 0, 129, 129, 102},
+ }
+ },{
+ {//FIXME 5/3
+ //8x8 dec=3
+ {275, 245, 245, 218},
+ { 0, 230, 230, 156},
+ { 0, 138, 138, 113},
+ },{
+ //16x16 dec=4
+ {352, 317, 317, 286},
+ { 0, 328, 328, 233},
+ { 0, 180, 180, 140},
+ { 0, 132, 132, 105},
+ }
+ }
+ };
+#endif
+
+ for (i = 0; i < h; i++) {
+ for (j = 0; j < w; j+=4) {
+ tmp[16*i+j+0] = (pix1[j+0] - pix2[j+0])<<4;
+ tmp[16*i+j+1] = (pix1[j+1] - pix2[j+1])<<4;
+ tmp[16*i+j+2] = (pix1[j+2] - pix2[j+2])<<4;
+ tmp[16*i+j+3] = (pix1[j+3] - pix2[j+3])<<4;
+ }
+ pix1 += line_size;
+ pix2 += line_size;
+ }
+ ff_spatial_dwt(tmp, w, h, 16, type, dec_count);
+
+ s=0;
+#if 0
+ for(level=0; level<dec_count; level++){
+ for(ori= level ? 1 : 0; ori<4; ori++){
+ int sx= (ori&1) ? 1<<level: 0;
+ int stride= 16<<(dec_count-level);
+ int sy= (ori&2) ? stride>>1 : 0;
+ int size= 1<<level;
+
+ for(i=0; i<size; i++){
+ for(j=0; j<size; j++){
+ int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
+ s += ABS(v);
+ }
+ }
+ }
+ }
+#endif
+ for (i = 0; i < h; i++) {
+ for (j = 0; j < w; j+=4) {
+ s+= ABS(tmp[16*i+j+0]);
+ s+= ABS(tmp[16*i+j+1]);
+ s+= ABS(tmp[16*i+j+2]);
+ s+= ABS(tmp[16*i+j+3]);
+ }
}
+ assert(s>=0);
+
+ return s>>2;
+}
+
+static int w53_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+ return w_c(v, pix1, pix2, line_size, 8, h, 1);
}
-void get_pixels_c(DCTELEM *block, const UINT8 *pixels, int line_size)
+static int w97_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+ return w_c(v, pix1, pix2, line_size, 8, h, 0);
+}
+
+static int w53_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+ return w_c(v, pix1, pix2, line_size, 16, h, 1);
+}
+
+static int w97_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+ return w_c(v, pix1, pix2, line_size, 16, h, 0);
+}
+
+static void get_pixels_c(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
{
- DCTELEM *p;
- const UINT8 *pix;
int i;
/* read the pixels */
- p = block;
- pix = pixels;
for(i=0;i<8;i++) {
- p[0] = pix[0];
- p[1] = pix[1];
- p[2] = pix[2];
- p[3] = pix[3];
- p[4] = pix[4];
- p[5] = pix[5];
- p[6] = pix[6];
- p[7] = pix[7];
- pix += line_size;
- p += 8;
+ block[0] = pixels[0];
+ block[1] = pixels[1];
+ block[2] = pixels[2];
+ block[3] = pixels[3];
+ block[4] = pixels[4];
+ block[5] = pixels[5];
+ block[6] = pixels[6];
+ block[7] = pixels[7];
+ pixels += line_size;
+ block += 8;
}
}
-void diff_pixels_c(DCTELEM *block, const UINT8 *s1, const UINT8 *s2, int stride){
- DCTELEM *p;
+static void diff_pixels_c(DCTELEM *restrict block, const uint8_t *s1,
+ const uint8_t *s2, int stride){
int i;
/* read the pixels */
- p = block;
for(i=0;i<8;i++) {
- p[0] = s1[0] - s2[0];
- p[1] = s1[1] - s2[1];
- p[2] = s1[2] - s2[2];
- p[3] = s1[3] - s2[3];
- p[4] = s1[4] - s2[4];
- p[5] = s1[5] - s2[5];
- p[6] = s1[6] - s2[6];
- p[7] = s1[7] - s2[7];
+ block[0] = s1[0] - s2[0];
+ block[1] = s1[1] - s2[1];
+ block[2] = s1[2] - s2[2];
+ block[3] = s1[3] - s2[3];
+ block[4] = s1[4] - s2[4];
+ block[5] = s1[5] - s2[5];
+ block[6] = s1[6] - s2[6];
+ block[7] = s1[7] - s2[7];
s1 += stride;
s2 += stride;
- p += 8;
+ block += 8;
}
}
-void put_pixels_clamped_c(const DCTELEM *block, UINT8 *pixels, int line_size)
+static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
{
- const DCTELEM *p;
- UINT8 *pix;
int i;
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
/* read the pixels */
- p = block;
- pix = pixels;
for(i=0;i<8;i++) {
- pix[0] = cm[p[0]];
- pix[1] = cm[p[1]];
- pix[2] = cm[p[2]];
- pix[3] = cm[p[3]];
- pix[4] = cm[p[4]];
- pix[5] = cm[p[5]];
- pix[6] = cm[p[6]];
- pix[7] = cm[p[7]];
- pix += line_size;
- p += 8;
+ pixels[0] = cm[block[0]];
+ pixels[1] = cm[block[1]];
+ pixels[2] = cm[block[2]];
+ pixels[3] = cm[block[3]];
+ pixels[4] = cm[block[4]];
+ pixels[5] = cm[block[5]];
+ pixels[6] = cm[block[6]];
+ pixels[7] = cm[block[7]];
+
+ pixels += line_size;
+ block += 8;
}
}
-void add_pixels_clamped_c(const DCTELEM *block, UINT8 *pixels, int line_size)
+static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
{
- const DCTELEM *p;
- UINT8 *pix;
int i;
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
/* read the pixels */
- p = block;
- pix = pixels;
- for(i=0;i<8;i++) {
- pix[0] = cm[pix[0] + p[0]];
- pix[1] = cm[pix[1] + p[1]];
- pix[2] = cm[pix[2] + p[2]];
- pix[3] = cm[pix[3] + p[3]];
- pix[4] = cm[pix[4] + p[4]];
- pix[5] = cm[pix[5] + p[5]];
- pix[6] = cm[pix[6] + p[6]];
- pix[7] = cm[pix[7] + p[7]];
- pix += line_size;
- p += 8;
+ for(i=0;i<4;i++) {
+ pixels[0] = cm[block[0]];
+ pixels[1] = cm[block[1]];
+ pixels[2] = cm[block[2]];
+ pixels[3] = cm[block[3]];
+
+ pixels += line_size;
+ block += 8;
}
}
-#ifdef __GNUC__
+static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
+{
+ int i;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ /* read the pixels */
+ for(i=0;i<2;i++) {
+ pixels[0] = cm[block[0]];
+ pixels[1] = cm[block[1]];
-struct unaligned_64 { uint64_t l; } __attribute__((packed));
-struct unaligned_32 { uint32_t l; } __attribute__((packed));
+ pixels += line_size;
+ block += 8;
+ }
+}
-#define LD32(a) (((const struct unaligned_32 *) (a))->l)
-#define LD64(a) (((const struct unaligned_64 *) (a))->l)
+static void put_signed_pixels_clamped_c(const DCTELEM *block,
+ uint8_t *restrict pixels,
+ int line_size)
+{
+ int i, j;
-#else /* __GNUC__ */
+ for (i = 0; i < 8; i++) {
+ for (j = 0; j < 8; j++) {
+ if (*block < -128)
+ *pixels = 0;
+ else if (*block > 127)
+ *pixels = 255;
+ else
+ *pixels = (uint8_t)(*block + 128);
+ block++;
+ pixels++;
+ }
+ pixels += (line_size - 8);
+ }
+}
-#define LD32(a) (*((uint32_t*)(a)))
-#define LD64(a) (*((uint64_t*)(a)))
+static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
+{
+ int i;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ /* read the pixels */
+ for(i=0;i<8;i++) {
+ pixels[0] = cm[pixels[0] + block[0]];
+ pixels[1] = cm[pixels[1] + block[1]];
+ pixels[2] = cm[pixels[2] + block[2]];
+ pixels[3] = cm[pixels[3] + block[3]];
+ pixels[4] = cm[pixels[4] + block[4]];
+ pixels[5] = cm[pixels[5] + block[5]];
+ pixels[6] = cm[pixels[6] + block[6]];
+ pixels[7] = cm[pixels[7] + block[7]];
+ pixels += line_size;
+ block += 8;
+ }
+}
-#endif /* !__GNUC__ */
+static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
+{
+ int i;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ /* read the pixels */
+ for(i=0;i<4;i++) {
+ pixels[0] = cm[pixels[0] + block[0]];
+ pixels[1] = cm[pixels[1] + block[1]];
+ pixels[2] = cm[pixels[2] + block[2]];
+ pixels[3] = cm[pixels[3] + block[3]];
+ pixels += line_size;
+ block += 8;
+ }
+}
+static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels,
+ int line_size)
+{
+ int i;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ /* read the pixels */
+ for(i=0;i<2;i++) {
+ pixels[0] = cm[pixels[0] + block[0]];
+ pixels[1] = cm[pixels[1] + block[1]];
+ pixels += line_size;
+ block += 8;
+ }
+}
#if 0
#define PIXOP2(OPNAME, OP) \
-void OPNAME ## _pixels(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static void OPNAME ## _pixels(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int i;\
for(i=0; i<h; i++){\
}\
}\
\
-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++){\
}\
}\
\
-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++){\
}\
}\
\
-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++){\
}\
}\
\
-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++){\
}\
}\
\
-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 );\
}\
}\
\
-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 );\
}\
}\
\
-void (*OPNAME ## _pixels_tab[4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- OPNAME ## _pixels,\
- OPNAME ## _pixels_x2,\
- OPNAME ## _pixels_y2,\
- OPNAME ## _pixels_xy2,\
-};\
-\
-void (*OPNAME ## _no_rnd_pixels_tab[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,\
-};
+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) \
-void OPNAME ## _pixels(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
-{\
+static void OPNAME ## _pixels2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ int i;\
+ for(i=0; i<h; i++){\
+ OP(*((uint16_t*)(block )), LD16(pixels ));\
+ pixels+=line_size;\
+ block +=line_size;\
+ }\
+}\
+static void OPNAME ## _pixels4_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 ));\
+ pixels+=line_size;\
+ block +=line_size;\
+ }\
+}\
+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_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels8_c(block, pixels, line_size, h);\
+}\
\
-void OPNAME ## _no_rnd_pixels_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
-{\
+static inline void OPNAME ## _no_rnd_pixels8_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
int i;\
for(i=0; i<h; i++){\
- int j;\
- for(j=0; j<2; j++){\
- const uint32_t a= LD32(pixels );\
- const uint32_t b= LD32(pixels+1);\
- OP(*((uint32_t*)block), (a&b) + (((a^b)&0xFEFEFEFEUL)>>1));\
- pixels+=4;\
- block +=4;\
- }\
- pixels+=line_size-8;\
- block +=line_size-8;\
+ uint32_t a,b;\
+ a= LD32(&src1[i*src_stride1 ]);\
+ b= LD32(&src2[i*src_stride2 ]);\
+ OP(*((uint32_t*)&dst[i*dst_stride ]), no_rnd_avg32(a, b));\
+ a= LD32(&src1[i*src_stride1+4]);\
+ b= LD32(&src2[i*src_stride2+4]);\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), no_rnd_avg32(a, b));\
}\
}\
\
-void OPNAME ## _pixels_x2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
-{\
+static inline void OPNAME ## _pixels8_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
int i;\
for(i=0; i<h; i++){\
- int j;\
- for(j=0; j<2; j++){\
- const uint32_t a= LD32(pixels );\
- const uint32_t b= LD32(pixels+1);\
- OP(*((uint32_t*)block), (a|b) - (((a^b)&0xFEFEFEFEUL)>>1));\
- pixels+=4;\
- block +=4;\
- }\
- pixels+=line_size-8;\
- block +=line_size-8;\
+ uint32_t a,b;\
+ a= LD32(&src1[i*src_stride1 ]);\
+ b= LD32(&src2[i*src_stride2 ]);\
+ OP(*((uint32_t*)&dst[i*dst_stride ]), rnd_avg32(a, b));\
+ a= LD32(&src1[i*src_stride1+4]);\
+ b= LD32(&src2[i*src_stride2+4]);\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), rnd_avg32(a, b));\
}\
}\
\
-void OPNAME ## _no_rnd_pixels_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
-{\
+static inline void OPNAME ## _pixels4_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
int i;\
for(i=0; i<h; i++){\
- int j;\
- for(j=0; j<2; j++){\
- const uint32_t a= LD32(pixels );\
- const uint32_t b= LD32(pixels+line_size);\
- OP(*((uint32_t*)block), (a&b) + (((a^b)&0xFEFEFEFEUL)>>1));\
- pixels+=4;\
- block +=4;\
- }\
- pixels+=line_size-8;\
- block +=line_size-8;\
+ uint32_t a,b;\
+ a= LD32(&src1[i*src_stride1 ]);\
+ b= LD32(&src2[i*src_stride2 ]);\
+ OP(*((uint32_t*)&dst[i*dst_stride ]), rnd_avg32(a, b));\
}\
}\
\
-void OPNAME ## _pixels_y2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
-{\
+static inline void OPNAME ## _pixels2_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
int i;\
for(i=0; i<h; i++){\
- int j;\
- for(j=0; j<2; j++){\
- const uint32_t a= LD32(pixels );\
- const uint32_t b= LD32(pixels+line_size);\
- OP(*((uint32_t*)block), (a|b) - (((a^b)&0xFEFEFEFEUL)>>1));\
- pixels+=4;\
- block +=4;\
- }\
- pixels+=line_size-8;\
- block +=line_size-8;\
+ uint32_t a,b;\
+ a= LD16(&src1[i*src_stride1 ]);\
+ b= LD16(&src2[i*src_stride2 ]);\
+ OP(*((uint16_t*)&dst[i*dst_stride ]), rnd_avg32(a, b));\
+ }\
+}\
+\
+static inline void OPNAME ## _pixels16_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
+ OPNAME ## _pixels8_l2(dst , src1 , src2 , dst_stride, src_stride1, src_stride2, h);\
+ OPNAME ## _pixels8_l2(dst+8, src1+8, src2+8, dst_stride, src_stride1, src_stride2, h);\
+}\
+\
+static inline void OPNAME ## _no_rnd_pixels16_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
+ int src_stride1, int src_stride2, int h){\
+ OPNAME ## _no_rnd_pixels8_l2(dst , src1 , src2 , dst_stride, src_stride1, src_stride2, h);\
+ 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_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_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_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_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);\
+}\
+\
+static inline void OPNAME ## _pixels8_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
+ int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
+ int i;\
+ for(i=0; i<h; i++){\
+ uint32_t a, b, c, d, l0, l1, h0, h1;\
+ a= LD32(&src1[i*src_stride1]);\
+ b= LD32(&src2[i*src_stride2]);\
+ c= LD32(&src3[i*src_stride3]);\
+ d= LD32(&src4[i*src_stride4]);\
+ l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x02020202UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ l1= (c&0x03030303UL)\
+ + (d&0x03030303UL);\
+ h1= ((c&0xFCFCFCFCUL)>>2)\
+ + ((d&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ a= LD32(&src1[i*src_stride1+4]);\
+ b= LD32(&src2[i*src_stride2+4]);\
+ c= LD32(&src3[i*src_stride3+4]);\
+ d= LD32(&src4[i*src_stride4+4]);\
+ l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x02020202UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ l1= (c&0x03030303UL)\
+ + (d&0x03030303UL);\
+ h1= ((c&0xFCFCFCFCUL)>>2)\
+ + ((d&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
}\
}\
\
-void OPNAME ## _pixels_xy2(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+static inline void OPNAME ## _pixels4_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels4_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
+}\
+\
+static inline void OPNAME ## _pixels4_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels4_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
+}\
+\
+static inline void OPNAME ## _pixels2_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels2_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
+}\
+\
+static inline void OPNAME ## _pixels2_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
+ OPNAME ## _pixels2_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
+}\
+\
+static inline void OPNAME ## _no_rnd_pixels8_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
+ int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
+ int i;\
+ for(i=0; i<h; i++){\
+ uint32_t a, b, c, d, l0, l1, h0, h1;\
+ a= LD32(&src1[i*src_stride1]);\
+ b= LD32(&src2[i*src_stride2]);\
+ c= LD32(&src3[i*src_stride3]);\
+ d= LD32(&src4[i*src_stride4]);\
+ l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x01010101UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ l1= (c&0x03030303UL)\
+ + (d&0x03030303UL);\
+ h1= ((c&0xFCFCFCFCUL)>>2)\
+ + ((d&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ a= LD32(&src1[i*src_stride1+4]);\
+ b= LD32(&src2[i*src_stride2+4]);\
+ c= LD32(&src3[i*src_stride3+4]);\
+ d= LD32(&src4[i*src_stride4+4]);\
+ l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x01010101UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ l1= (c&0x03030303UL)\
+ + (d&0x03030303UL);\
+ h1= ((c&0xFCFCFCFCUL)>>2)\
+ + ((d&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ }\
+}\
+static inline void OPNAME ## _pixels16_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
+ int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
+ OPNAME ## _pixels8_l4(dst , src1 , src2 , src3 , src4 , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
+ OPNAME ## _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 ## _no_rnd_pixels16_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
+ int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
+ OPNAME ## _no_rnd_pixels8_l4(dst , src1 , src2 , src3 , src4 , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, 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 ## _pixels2_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
+{\
+ int i, a0, b0, a1, b1;\
+ a0= pixels[0];\
+ b0= pixels[1] + 2;\
+ a0 += b0;\
+ b0 += pixels[2];\
+\
+ pixels+=line_size;\
+ for(i=0; i<h; i+=2){\
+ a1= pixels[0];\
+ b1= pixels[1];\
+ a1 += b1;\
+ b1 += pixels[2];\
+\
+ block[0]= (a1+a0)>>2; /* FIXME non put */\
+ block[1]= (b1+b0)>>2;\
+\
+ pixels+=line_size;\
+ block +=line_size;\
+\
+ a0= pixels[0];\
+ b0= pixels[1] + 2;\
+ a0 += b0;\
+ b0 += pixels[2];\
+\
+ block[0]= (a1+a0)>>2;\
+ block[1]= (b1+b0)>>2;\
+ pixels+=line_size;\
+ block +=line_size;\
+ }\
+}\
+\
+static inline void OPNAME ## _pixels4_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
- int j;\
- for(j=0; j<2; j++){\
int i;\
const uint32_t a= LD32(pixels );\
const uint32_t b= LD32(pixels+1);\
pixels+=line_size;\
block +=line_size;\
}\
- pixels+=4-line_size*(h+1);\
- block +=4-line_size*h;\
- }\
}\
\
-void OPNAME ## _no_rnd_pixels_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++){\
const uint32_t b= LD32(pixels+1);\
uint32_t l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
- + 0x01010101UL;\
+ + 0x02020202UL;\
uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
uint32_t l1,h1;\
b= LD32(pixels+1);\
l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
- + 0x01010101UL;\
+ + 0x02020202UL;\
h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
}\
}\
\
-void (*OPNAME ## _pixels_tab[4])(uint8_t *block, const uint8_t *pixels, int line_size, int h) = {\
- OPNAME ## _pixels,\
- OPNAME ## _pixels_x2,\
- OPNAME ## _pixels_y2,\
- OPNAME ## _pixels_xy2,\
-};\
+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++){\
+ int i;\
+ const uint32_t a= LD32(pixels );\
+ const uint32_t b= LD32(pixels+1);\
+ uint32_t l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x01010101UL;\
+ uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ uint32_t l1,h1;\
\
-void (*OPNAME ## _no_rnd_pixels_tab[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,\
-};
-#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEUL)>>1) )
-#endif
-
-#define op_put(a, b) a = b
-
-PIXOP2(avg, op_avg)
-PIXOP2(put, op_put)
-#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
-
-PIXOP(DCTELEM, sub, op_sub, 8)
-
-/* 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)
-
-/* motion estimation */
-
-#undef avg2
-#undef avg4
+ pixels+=line_size;\
+ for(i=0; i<h; i+=2){\
+ uint32_t a= LD32(pixels );\
+ uint32_t b= LD32(pixels+1);\
+ l1= (a&0x03030303UL)\
+ + (b&0x03030303UL);\
+ h1= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ pixels+=line_size;\
+ block +=line_size;\
+ a= LD32(pixels );\
+ b= LD32(pixels+1);\
+ l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x01010101UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ pixels+=line_size;\
+ block +=line_size;\
+ }\
+ pixels+=4-line_size*(h+1);\
+ block +=4-line_size*h;\
+ }\
+}\
+\
+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 = rnd_avg32(a, b)
#endif
+#define op_put(a, b) a = b
+
+PIXOP2(avg, op_avg)
+PIXOP2(put, op_put)
+#undef op_avg
+#undef op_put
#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 srcStride, int h, int x16, int y16, int rounder)
+static void put_no_rnd_pixels16_l2_c(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
+ put_no_rnd_pixels16_l2(dst, a, b, stride, stride, stride, h);
+}
+
+static void put_no_rnd_pixels8_l2_c(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){
+ put_no_rnd_pixels8_l2(dst, a, b, stride, stride, stride, h);
+}
+
+static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y16, int rounder)
{
const int A=(16-x16)*(16-y16);
const int B=( x16)*(16-y16);
const int C=(16-x16)*( y16);
const int D=( x16)*( y16);
int i;
- rounder= 128 - rounder;
for(i=0; i<h; i++)
{
- dst[0]= (A*src[0] + B*src[1] + C*src[srcStride+0] + D*src[srcStride+1] + rounder)>>8;
- dst[1]= (A*src[1] + B*src[2] + C*src[srcStride+1] + D*src[srcStride+2] + rounder)>>8;
- dst[2]= (A*src[2] + B*src[3] + C*src[srcStride+2] + D*src[srcStride+3] + rounder)>>8;
- dst[3]= (A*src[3] + B*src[4] + C*src[srcStride+3] + D*src[srcStride+4] + rounder)>>8;
- dst[4]= (A*src[4] + B*src[5] + C*src[srcStride+4] + D*src[srcStride+5] + rounder)>>8;
- dst[5]= (A*src[5] + B*src[6] + C*src[srcStride+5] + D*src[srcStride+6] + rounder)>>8;
- dst[6]= (A*src[6] + B*src[7] + C*src[srcStride+6] + D*src[srcStride+7] + rounder)>>8;
- dst[7]= (A*src[7] + B*src[8] + C*src[srcStride+7] + D*src[srcStride+8] + rounder)>>8;
- dst+= srcStride;
- src+= srcStride;
+ dst[0]= (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + rounder)>>8;
+ dst[1]= (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + rounder)>>8;
+ dst[2]= (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + rounder)>>8;
+ dst[3]= (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + rounder)>>8;
+ dst[4]= (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + rounder)>>8;
+ dst[5]= (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + rounder)>>8;
+ dst[6]= (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + rounder)>>8;
+ dst[7]= (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + rounder)>>8;
+ dst+= stride;
+ src+= stride;
+ }
+}
+
+static void gmc_c(uint8_t *dst, uint8_t *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 put_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ switch(width){
+ case 2: put_pixels2_c (dst, src, stride, height); break;
+ case 4: put_pixels4_c (dst, src, stride, height); break;
+ case 8: put_pixels8_c (dst, src, stride, height); break;
+ case 16:put_pixels16_c(dst, src, stride, height); break;
+ }
+}
+
+static inline void put_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (683*(2*src[j] + src[j+1] + 1)) >> 11;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (683*(src[j] + 2*src[j+1] + 1)) >> 11;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (683*(2*src[j] + src[j+stride] + 1)) >> 11;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (683*(src[j] + 2*src[j+stride] + 1)) >> 11;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ switch(width){
+ case 2: avg_pixels2_c (dst, src, stride, height); break;
+ case 4: avg_pixels4_c (dst, src, stride, height); break;
+ case 8: avg_pixels8_c (dst, src, stride, height); break;
+ case 16:avg_pixels16_c(dst, src, stride, height); break;
+ }
+}
+
+static inline void avg_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((683*(2*src[j] + src[j+1] + 1)) >> 11) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+1] + 1)) >> 11) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((683*(2*src[j] + src[j+stride] + 1)) >> 11) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
+ }
+}
+
+static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+ int i,j;
+ for (i=0; i < height; i++) {
+ for (j=0; j < width; j++) {
+ dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+ }
+ src += stride;
+ dst += stride;
}
}
+#if 0
+#define TPEL_WIDTH(width)\
+static void put_tpel_pixels ## width ## _mc00_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc00_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc10_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc10_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc20_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc20_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc01_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc01_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc11_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc11_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc21_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc21_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc02_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc02_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc12_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc12_c(dst, src, stride, width, height);}\
+static void put_tpel_pixels ## width ## _mc22_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
+ void put_tpel_pixels_mc22_c(dst, src, stride, width, height);}
+#endif
-static void qpel_h_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h, int r)
+#define H264_CHROMA_MC(OPNAME, OP)\
+static void OPNAME ## h264_chroma_mc2_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
+ const int A=(8-x)*(8-y);\
+ const int B=( x)*(8-y);\
+ const int C=(8-x)*( y);\
+ const int D=( x)*( y);\
+ int i;\
+ \
+ assert(x<8 && y<8 && x>=0 && y>=0);\
+\
+ for(i=0; i<h; i++)\
+ {\
+ OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
+ OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
+ dst+= stride;\
+ src+= stride;\
+ }\
+}\
+\
+static void OPNAME ## h264_chroma_mc4_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
+ const int A=(8-x)*(8-y);\
+ const int B=( x)*(8-y);\
+ const int C=(8-x)*( y);\
+ const int D=( x)*( y);\
+ int i;\
+ \
+ assert(x<8 && y<8 && x>=0 && y>=0);\
+\
+ for(i=0; i<h; i++)\
+ {\
+ OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
+ OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
+ OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
+ OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
+ dst+= stride;\
+ src+= stride;\
+ }\
+}\
+\
+static void OPNAME ## h264_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
+ const int A=(8-x)*(8-y);\
+ const int B=( x)*(8-y);\
+ const int C=(8-x)*( y);\
+ const int D=( x)*( y);\
+ int i;\
+ \
+ assert(x<8 && y<8 && x>=0 && y>=0);\
+\
+ for(i=0; i<h; i++)\
+ {\
+ OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
+ OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2]));\
+ OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3]));\
+ OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4]));\
+ OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5]));\
+ OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6]));\
+ OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7]));\
+ OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8]));\
+ dst+= stride;\
+ src+= stride;\
+ }\
+}
+
+#define op_avg(a, b) a = (((a)+(((b) + 32)>>6)+1)>>1)
+#define op_put(a, b) a = (((b) + 32)>>6)
+
+H264_CHROMA_MC(put_ , op_put)
+H264_CHROMA_MC(avg_ , op_avg)
+#undef op_avg
+#undef op_put
+
+static inline void copy_block4(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
int i;
for(i=0; i<h; i++)
{
- dst[0]= cm[(((src[0]+src[1])*20 - (src[0]+src[2])*6 + (src[1]+src[3])*3 - (src[2]+src[4]) + r)>>5)];
- dst[1]= cm[(((src[1]+src[2])*20 - (src[0]+src[3])*6 + (src[0]+src[4])*3 - (src[1]+src[5]) + r)>>5)];
- dst[2]= cm[(((src[2]+src[3])*20 - (src[1]+src[4])*6 + (src[0]+src[5])*3 - (src[0]+src[6]) + r)>>5)];
- dst[3]= cm[(((src[3]+src[4])*20 - (src[2]+src[5])*6 + (src[1]+src[6])*3 - (src[0]+src[7]) + r)>>5)];
- dst[4]= cm[(((src[4]+src[5])*20 - (src[3]+src[6])*6 + (src[2]+src[7])*3 - (src[1]+src[8]) + r)>>5)];
- dst[5]= cm[(((src[5]+src[6])*20 - (src[4]+src[7])*6 + (src[3]+src[8])*3 - (src[2]+src[8]) + r)>>5)];
- dst[6]= cm[(((src[6]+src[7])*20 - (src[5]+src[8])*6 + (src[4]+src[8])*3 - (src[3]+src[7]) + r)>>5)];
- dst[7]= cm[(((src[7]+src[8])*20 - (src[6]+src[8])*6 + (src[5]+src[7])*3 - (src[4]+src[6]) + r)>>5)];
+ ST32(dst , LD32(src ));
dst+=dstStride;
src+=srcStride;
}
}
-static void qpel_v_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int w, int r)
+static inline void copy_block8(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
int i;
- for(i=0; i<w; i++)
+ for(i=0; i<h; i++)
{
- const int src0= src[0*srcStride];
- const int src1= src[1*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];
- dst[0*dstStride]= cm[(((src0+src1)*20 - (src0+src2)*6 + (src1+src3)*3 - (src2+src4) + r)>>5)];
- dst[1*dstStride]= cm[(((src1+src2)*20 - (src0+src3)*6 + (src0+src4)*3 - (src1+src5) + r)>>5)];
- dst[2*dstStride]= cm[(((src2+src3)*20 - (src1+src4)*6 + (src0+src5)*3 - (src0+src6) + r)>>5)];
- dst[3*dstStride]= cm[(((src3+src4)*20 - (src2+src5)*6 + (src1+src6)*3 - (src0+src7) + r)>>5)];
- dst[4*dstStride]= cm[(((src4+src5)*20 - (src3+src6)*6 + (src2+src7)*3 - (src1+src8) + r)>>5)];
- dst[5*dstStride]= cm[(((src5+src6)*20 - (src4+src7)*6 + (src3+src8)*3 - (src2+src8) + r)>>5)];
- dst[6*dstStride]= cm[(((src6+src7)*20 - (src5+src8)*6 + (src4+src8)*3 - (src3+src7) + r)>>5)];
- dst[7*dstStride]= cm[(((src7+src8)*20 - (src6+src8)*6 + (src5+src7)*3 - (src4+src6) + r)>>5)];
- dst++;
- src++;
+ ST32(dst , LD32(src ));
+ ST32(dst+4 , LD32(src+4 ));
+ dst+=dstStride;
+ src+=srcStride;
}
}
-static inline void put_block(UINT8 *dst, UINT8 *src, int dstStride, int srcStride)
+static inline void copy_block16(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
int i;
- for(i=0; i<8; i++)
+ for(i=0; i<h; i++)
{
- dst[0]= src[0];
- dst[1]= src[1];
- dst[2]= src[2];
- dst[3]= src[3];
- dst[4]= src[4];
- dst[5]= src[5];
- dst[6]= src[6];
- dst[7]= src[7];
+ ST32(dst , LD32(src ));
+ ST32(dst+4 , LD32(src+4 ));
+ ST32(dst+8 , LD32(src+8 ));
+ ST32(dst+12, LD32(src+12));
dst+=dstStride;
src+=srcStride;
}
}
-static inline void avg2_block(UINT8 *dst, UINT8 *src1, UINT8 *src2, int dstStride, int srcStride, int r)
+static inline void copy_block17(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
int i;
- for(i=0; i<8; i++)
+ for(i=0; i<h; i++)
{
- dst[0]= (src1[0] + src2[0] + r)>>1;
- dst[1]= (src1[1] + src2[1] + r)>>1;
- dst[2]= (src1[2] + src2[2] + r)>>1;
- dst[3]= (src1[3] + src2[3] + r)>>1;
- dst[4]= (src1[4] + src2[4] + r)>>1;
- dst[5]= (src1[5] + src2[5] + r)>>1;
- dst[6]= (src1[6] + src2[6] + r)>>1;
- dst[7]= (src1[7] + src2[7] + r)>>1;
+ ST32(dst , LD32(src ));
+ ST32(dst+4 , LD32(src+4 ));
+ ST32(dst+8 , LD32(src+8 ));
+ ST32(dst+12, LD32(src+12));
+ dst[16]= src[16];
dst+=dstStride;
- src1+=srcStride;
- src2+=8;
+ src+=srcStride;
}
}
-static inline void avg4_block(UINT8 *dst, UINT8 *src1, UINT8 *src2, UINT8 *src3, UINT8 *src4, int dstStride, int srcStride, int r)
+static inline void copy_block9(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
int i;
- for(i=0; i<8; i++)
+ for(i=0; i<h; i++)
{
- dst[0]= (src1[0] + src2[0] + src3[0] + src4[0] + r)>>2;
- dst[1]= (src1[1] + src2[1] + src3[1] + src4[1] + r)>>2;
- dst[2]= (src1[2] + src2[2] + src3[2] + src4[2] + r)>>2;
- dst[3]= (src1[3] + src2[3] + src3[3] + src4[3] + r)>>2;
- dst[4]= (src1[4] + src2[4] + src3[4] + src4[4] + r)>>2;
- dst[5]= (src1[5] + src2[5] + src3[5] + src4[5] + r)>>2;
- dst[6]= (src1[6] + src2[6] + src3[6] + src4[6] + r)>>2;
- dst[7]= (src1[7] + src2[7] + src3[7] + src4[7] + r)>>2;
+ ST32(dst , LD32(src ));
+ ST32(dst+4 , LD32(src+4 ));
+ dst[8]= src[8];
+ dst+=dstStride;
+ src+=srcStride;
+ }
+}
+
+
+#define QPEL_MC(r, OPNAME, RND, OP) \
+static void OPNAME ## mpeg4_qpel8_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++)\
+ {\
+ OP(dst[0], (src[0]+src[1])*20 - (src[0]+src[2])*6 + (src[1]+src[3])*3 - (src[2]+src[4]));\
+ OP(dst[1], (src[1]+src[2])*20 - (src[0]+src[3])*6 + (src[0]+src[4])*3 - (src[1]+src[5]));\
+ OP(dst[2], (src[2]+src[3])*20 - (src[1]+src[4])*6 + (src[0]+src[5])*3 - (src[0]+src[6]));\
+ OP(dst[3], (src[3]+src[4])*20 - (src[2]+src[5])*6 + (src[1]+src[6])*3 - (src[0]+src[7]));\
+ OP(dst[4], (src[4]+src[5])*20 - (src[3]+src[6])*6 + (src[2]+src[7])*3 - (src[1]+src[8]));\
+ OP(dst[5], (src[5]+src[6])*20 - (src[4]+src[7])*6 + (src[3]+src[8])*3 - (src[2]+src[8]));\
+ OP(dst[6], (src[6]+src[7])*20 - (src[5]+src[8])*6 + (src[4]+src[8])*3 - (src[3]+src[7]));\
+ OP(dst[7], (src[7]+src[8])*20 - (src[6]+src[8])*6 + (src[5]+src[7])*3 - (src[4]+src[6]));\
+ dst+=dstStride;\
+ src+=srcStride;\
+ }\
+}\
+\
+static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ const int w=8;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ for(i=0; i<w; i++)\
+ {\
+ const int src0= src[0*srcStride];\
+ const int src1= src[1*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];\
+ OP(dst[0*dstStride], (src0+src1)*20 - (src0+src2)*6 + (src1+src3)*3 - (src2+src4));\
+ OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*6 + (src0+src4)*3 - (src1+src5));\
+ OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*6 + (src0+src5)*3 - (src0+src6));\
+ OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*6 + (src1+src6)*3 - (src0+src7));\
+ OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*6 + (src2+src7)*3 - (src1+src8));\
+ OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*6 + (src3+src8)*3 - (src2+src8));\
+ OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*6 + (src4+src8)*3 - (src3+src7));\
+ OP(dst[7*dstStride], (src7+src8)*20 - (src6+src8)*6 + (src5+src7)*3 - (src4+src6));\
+ dst++;\
+ src++;\
+ }\
+}\
+\
+static void OPNAME ## mpeg4_qpel16_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++)\
+ {\
+ OP(dst[ 0], (src[ 0]+src[ 1])*20 - (src[ 0]+src[ 2])*6 + (src[ 1]+src[ 3])*3 - (src[ 2]+src[ 4]));\
+ OP(dst[ 1], (src[ 1]+src[ 2])*20 - (src[ 0]+src[ 3])*6 + (src[ 0]+src[ 4])*3 - (src[ 1]+src[ 5]));\
+ OP(dst[ 2], (src[ 2]+src[ 3])*20 - (src[ 1]+src[ 4])*6 + (src[ 0]+src[ 5])*3 - (src[ 0]+src[ 6]));\
+ OP(dst[ 3], (src[ 3]+src[ 4])*20 - (src[ 2]+src[ 5])*6 + (src[ 1]+src[ 6])*3 - (src[ 0]+src[ 7]));\
+ OP(dst[ 4], (src[ 4]+src[ 5])*20 - (src[ 3]+src[ 6])*6 + (src[ 2]+src[ 7])*3 - (src[ 1]+src[ 8]));\
+ OP(dst[ 5], (src[ 5]+src[ 6])*20 - (src[ 4]+src[ 7])*6 + (src[ 3]+src[ 8])*3 - (src[ 2]+src[ 9]));\
+ OP(dst[ 6], (src[ 6]+src[ 7])*20 - (src[ 5]+src[ 8])*6 + (src[ 4]+src[ 9])*3 - (src[ 3]+src[10]));\
+ OP(dst[ 7], (src[ 7]+src[ 8])*20 - (src[ 6]+src[ 9])*6 + (src[ 5]+src[10])*3 - (src[ 4]+src[11]));\
+ OP(dst[ 8], (src[ 8]+src[ 9])*20 - (src[ 7]+src[10])*6 + (src[ 6]+src[11])*3 - (src[ 5]+src[12]));\
+ OP(dst[ 9], (src[ 9]+src[10])*20 - (src[ 8]+src[11])*6 + (src[ 7]+src[12])*3 - (src[ 6]+src[13]));\
+ OP(dst[10], (src[10]+src[11])*20 - (src[ 9]+src[12])*6 + (src[ 8]+src[13])*3 - (src[ 7]+src[14]));\
+ OP(dst[11], (src[11]+src[12])*20 - (src[10]+src[13])*6 + (src[ 9]+src[14])*3 - (src[ 8]+src[15]));\
+ OP(dst[12], (src[12]+src[13])*20 - (src[11]+src[14])*6 + (src[10]+src[15])*3 - (src[ 9]+src[16]));\
+ OP(dst[13], (src[13]+src[14])*20 - (src[12]+src[15])*6 + (src[11]+src[16])*3 - (src[10]+src[16]));\
+ OP(dst[14], (src[14]+src[15])*20 - (src[13]+src[16])*6 + (src[12]+src[16])*3 - (src[11]+src[15]));\
+ OP(dst[15], (src[15]+src[16])*20 - (src[14]+src[16])*6 + (src[13]+src[15])*3 - (src[12]+src[14]));\
+ dst+=dstStride;\
+ src+=srcStride;\
+ }\
+}\
+\
+static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ const int w=16;\
+ for(i=0; i<w; i++)\
+ {\
+ const int src0= src[0*srcStride];\
+ const int src1= src[1*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];\
+ const int src10= src[10*srcStride];\
+ const int src11= src[11*srcStride];\
+ const int src12= src[12*srcStride];\
+ const int src13= src[13*srcStride];\
+ const int src14= src[14*srcStride];\
+ const int src15= src[15*srcStride];\
+ const int src16= src[16*srcStride];\
+ OP(dst[ 0*dstStride], (src0 +src1 )*20 - (src0 +src2 )*6 + (src1 +src3 )*3 - (src2 +src4 ));\
+ OP(dst[ 1*dstStride], (src1 +src2 )*20 - (src0 +src3 )*6 + (src0 +src4 )*3 - (src1 +src5 ));\
+ OP(dst[ 2*dstStride], (src2 +src3 )*20 - (src1 +src4 )*6 + (src0 +src5 )*3 - (src0 +src6 ));\
+ OP(dst[ 3*dstStride], (src3 +src4 )*20 - (src2 +src5 )*6 + (src1 +src6 )*3 - (src0 +src7 ));\
+ OP(dst[ 4*dstStride], (src4 +src5 )*20 - (src3 +src6 )*6 + (src2 +src7 )*3 - (src1 +src8 ));\
+ OP(dst[ 5*dstStride], (src5 +src6 )*20 - (src4 +src7 )*6 + (src3 +src8 )*3 - (src2 +src9 ));\
+ OP(dst[ 6*dstStride], (src6 +src7 )*20 - (src5 +src8 )*6 + (src4 +src9 )*3 - (src3 +src10));\
+ OP(dst[ 7*dstStride], (src7 +src8 )*20 - (src6 +src9 )*6 + (src5 +src10)*3 - (src4 +src11));\
+ OP(dst[ 8*dstStride], (src8 +src9 )*20 - (src7 +src10)*6 + (src6 +src11)*3 - (src5 +src12));\
+ OP(dst[ 9*dstStride], (src9 +src10)*20 - (src8 +src11)*6 + (src7 +src12)*3 - (src6 +src13));\
+ OP(dst[10*dstStride], (src10+src11)*20 - (src9 +src12)*6 + (src8 +src13)*3 - (src7 +src14));\
+ OP(dst[11*dstStride], (src11+src12)*20 - (src10+src13)*6 + (src9 +src14)*3 - (src8 +src15));\
+ OP(dst[12*dstStride], (src12+src13)*20 - (src11+src14)*6 + (src10+src15)*3 - (src9 +src16));\
+ OP(dst[13*dstStride], (src13+src14)*20 - (src12+src15)*6 + (src11+src16)*3 - (src10+src16));\
+ OP(dst[14*dstStride], (src14+src15)*20 - (src13+src16)*6 + (src12+src16)*3 - (src11+src15));\
+ OP(dst[15*dstStride], (src15+src16)*20 - (src14+src16)*6 + (src13+src15)*3 - (src12+src14));\
+ dst++;\
+ src++;\
+ }\
+}\
+\
+static void OPNAME ## qpel8_mc00_c (uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## pixels8_c(dst, src, stride, 8);\
+}\
+\
+static void OPNAME ## qpel8_mc10_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[64];\
+ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\
+ OPNAME ## pixels8_l2(dst, src, half, stride, stride, 8, 8);\
+}\
+\
+static void OPNAME ## qpel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## mpeg4_qpel8_h_lowpass(dst, src, stride, stride, 8);\
+}\
+\
+static void OPNAME ## qpel8_mc30_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[64];\
+ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\
+ OPNAME ## pixels8_l2(dst, src+1, half, stride, stride, 8, 8);\
+}\
+\
+static void OPNAME ## qpel8_mc01_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t half[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\
+ OPNAME ## pixels8_l2(dst, full, half, stride, 16, 8, 8);\
+}\
+\
+static void OPNAME ## qpel8_mc02_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ copy_block9(full, src, 16, stride, 9);\
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, full, stride, 16);\
+}\
+\
+static void OPNAME ## qpel8_mc03_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t half[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\
+ OPNAME ## pixels8_l2(dst, full+16, half, stride, 16, 8, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l4(dst, full, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc11_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l4(dst, full+1, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc31_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l4(dst, full+16, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc13_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full , 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l4(dst, full+17, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc21_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc23_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[72];\
+ uint8_t halfHV[64];\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc12_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc12_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
+}\
+void ff_ ## OPNAME ## qpel8_mc32_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ uint8_t halfV[64];\
+ uint8_t halfHV[64];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\
+ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
+ OPNAME ## pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);\
+}\
+static void OPNAME ## qpel8_mc32_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[16*9];\
+ uint8_t halfH[72];\
+ copy_block9(full, src, 16, stride, 9);\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
+ put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
+}\
+static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[72];\
+ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
+ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
+}\
+static void OPNAME ## qpel16_mc00_c (uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## pixels16_c(dst, src, stride, 16);\
+}\
+\
+static void OPNAME ## qpel16_mc10_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[256];\
+ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\
+ OPNAME ## pixels16_l2(dst, src, half, stride, stride, 16, 16);\
+}\
+\
+static void OPNAME ## qpel16_mc20_c(uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## mpeg4_qpel16_h_lowpass(dst, src, stride, stride, 16);\
+}\
+\
+static void OPNAME ## qpel16_mc30_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[256];\
+ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\
+ OPNAME ## pixels16_l2(dst, src+1, half, stride, stride, 16, 16);\
+}\
+\
+static void OPNAME ## qpel16_mc01_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t half[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\
+ OPNAME ## pixels16_l2(dst, full, half, stride, 24, 16, 16);\
+}\
+\
+static void OPNAME ## qpel16_mc02_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ copy_block17(full, src, 24, stride, 17);\
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, full, stride, 24);\
+}\
+\
+static void OPNAME ## qpel16_mc03_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t half[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\
+ OPNAME ## pixels16_l2(dst, full+24, half, stride, 24, 16, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l4(dst, full, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc11_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l4(dst, full+1, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc31_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l4(dst, full+24, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc13_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full , 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l4(dst, full+25, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc21_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc23_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[272];\
+ uint8_t halfHV[256];\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc12_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfV, halfHV, stride, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc12_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
+}\
+void ff_ ## OPNAME ## qpel16_mc32_old_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ uint8_t halfV[256];\
+ uint8_t halfHV[256];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\
+ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
+ OPNAME ## pixels16_l2(dst, halfV, halfHV, stride, 16, 16, 16);\
+}\
+static void OPNAME ## qpel16_mc32_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[24*17];\
+ uint8_t halfH[272];\
+ copy_block17(full, src, 24, stride, 17);\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
+ put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
+}\
+static void OPNAME ## qpel16_mc22_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t halfH[272];\
+ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
+ OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
+}
+
+#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)
+#define op_put(a, b) a = cm[((b) + 16)>>5]
+#define op_put_no_rnd(a, b) a = cm[((b) + 15)>>5]
+
+QPEL_MC(0, put_ , _ , op_put)
+QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd)
+QPEL_MC(0, avg_ , _ , op_avg)
+//QPEL_MC(1, avg_no_rnd , _ , op_avg)
+#undef op_avg
+#undef op_avg_no_rnd
+#undef op_put
+#undef op_put_no_rnd
+
+#if 1
+#define H264_LOWPASS(OPNAME, OP, OP2) \
+static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ const int h=4;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ for(i=0; i<h; i++)\
+ {\
+ OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
+ OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
+ OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\
+ OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\
+ dst+=dstStride;\
+ src+=srcStride;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel4_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ const int w=4;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ for(i=0; i<w; i++)\
+ {\
+ const int srcB= src[-2*srcStride];\
+ const int srcA= src[-1*srcStride];\
+ const int src0= src[0 *srcStride];\
+ const int src1= src[1 *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];\
+ OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
+ OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
+ OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
+ OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
+ dst++;\
+ src++;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel4_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+ const int h=4;\
+ const int w=4;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ src -= 2*srcStride;\
+ for(i=0; i<h+5; i++)\
+ {\
+ tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]);\
+ tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]);\
+ tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]);\
+ tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]);\
+ tmp+=tmpStride;\
+ src+=srcStride;\
+ }\
+ tmp -= tmpStride*(h+5-2);\
+ for(i=0; i<w; i++)\
+ {\
+ const int tmpB= tmp[-2*tmpStride];\
+ const int tmpA= tmp[-1*tmpStride];\
+ const int tmp0= tmp[0 *tmpStride];\
+ const int tmp1= tmp[1 *tmpStride];\
+ const int tmp2= tmp[2 *tmpStride];\
+ const int tmp3= tmp[3 *tmpStride];\
+ const int tmp4= tmp[4 *tmpStride];\
+ const int tmp5= tmp[5 *tmpStride];\
+ const int tmp6= tmp[6 *tmpStride];\
+ OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
+ OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
+ OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
+ OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
+ dst++;\
+ tmp++;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ const int h=8;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ for(i=0; i<h; i++)\
+ {\
+ OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\
+ OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\
+ OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\
+ OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\
+ OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\
+ OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\
+ OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\
+ OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\
+ dst+=dstStride;\
+ src+=srcStride;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ const int w=8;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ for(i=0; i<w; i++)\
+ {\
+ const int srcB= src[-2*srcStride];\
+ const int srcA= src[-1*srcStride];\
+ const int src0= src[0 *srcStride];\
+ const int src1= src[1 *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];\
+ const int src10=src[10*srcStride];\
+ OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
+ OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
+ OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
+ OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
+ OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
+ OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
+ OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
+ OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
+ dst++;\
+ src++;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel8_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+ const int h=8;\
+ const int w=8;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+ int i;\
+ src -= 2*srcStride;\
+ for(i=0; i<h+5; i++)\
+ {\
+ tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]);\
+ tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]);\
+ tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]);\
+ tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]);\
+ tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]);\
+ tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]);\
+ tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]);\
+ tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]);\
+ tmp+=tmpStride;\
+ src+=srcStride;\
+ }\
+ tmp -= tmpStride*(h+5-2);\
+ for(i=0; i<w; i++)\
+ {\
+ const int tmpB= tmp[-2*tmpStride];\
+ const int tmpA= tmp[-1*tmpStride];\
+ const int tmp0= tmp[0 *tmpStride];\
+ const int tmp1= tmp[1 *tmpStride];\
+ const int tmp2= tmp[2 *tmpStride];\
+ const int tmp3= tmp[3 *tmpStride];\
+ const int tmp4= tmp[4 *tmpStride];\
+ const int tmp5= tmp[5 *tmpStride];\
+ const int tmp6= tmp[6 *tmpStride];\
+ const int tmp7= tmp[7 *tmpStride];\
+ const int tmp8= tmp[8 *tmpStride];\
+ const int tmp9= tmp[9 *tmpStride];\
+ const int tmp10=tmp[10*tmpStride];\
+ OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
+ OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
+ OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
+ OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
+ OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\
+ OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\
+ OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\
+ OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\
+ dst++;\
+ tmp++;\
+ }\
+}\
+\
+static void OPNAME ## h264_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ OPNAME ## h264_qpel8_v_lowpass(dst , src , dstStride, srcStride);\
+ OPNAME ## h264_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride);\
+ src += 8*srcStride;\
+ dst += 8*dstStride;\
+ OPNAME ## h264_qpel8_v_lowpass(dst , src , dstStride, srcStride);\
+ OPNAME ## h264_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride);\
+}\
+\
+static void OPNAME ## h264_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+ OPNAME ## h264_qpel8_h_lowpass(dst , src , dstStride, srcStride);\
+ OPNAME ## h264_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride);\
+ src += 8*srcStride;\
+ dst += 8*dstStride;\
+ OPNAME ## h264_qpel8_h_lowpass(dst , src , dstStride, srcStride);\
+ OPNAME ## h264_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride);\
+}\
+\
+static void OPNAME ## h264_qpel16_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+ OPNAME ## h264_qpel8_hv_lowpass(dst , tmp , src , dstStride, tmpStride, srcStride);\
+ OPNAME ## h264_qpel8_hv_lowpass(dst+8, tmp+8, src+8, dstStride, tmpStride, srcStride);\
+ src += 8*srcStride;\
+ dst += 8*dstStride;\
+ OPNAME ## h264_qpel8_hv_lowpass(dst , tmp , src , dstStride, tmpStride, srcStride);\
+ OPNAME ## h264_qpel8_hv_lowpass(dst+8, tmp+8, src+8, dstStride, tmpStride, srcStride);\
+}\
+
+#define H264_MC(OPNAME, SIZE) \
+static void OPNAME ## h264_qpel ## SIZE ## _mc00_c (uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## pixels ## SIZE ## _c(dst, src, stride, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, src, half, stride, stride, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
+ OPNAME ## h264_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t half[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, src+1, half, stride, stride, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t half[SIZE*SIZE];\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, full_mid, half, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ OPNAME ## h264_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t half[SIZE*SIZE];\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, full_mid+SIZE, half, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
+ copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc33_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
+ copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
+ int16_t tmp[SIZE*(SIZE+5)];\
+ OPNAME ## h264_qpel ## SIZE ## _hv_lowpass(dst, tmp, src, stride, SIZE, stride);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
+ int16_t tmp[SIZE*(SIZE+5)];\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfHV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
+ put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, int stride){\
+ int16_t tmp[SIZE*(SIZE+5)];\
+ uint8_t halfH[SIZE*SIZE];\
+ uint8_t halfHV[SIZE*SIZE];\
+ put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
+ put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ int16_t tmp[SIZE*(SIZE+5)];\
+ uint8_t halfV[SIZE*SIZE];\
+ uint8_t halfHV[SIZE*SIZE];\
+ copy_block ## SIZE (full, src - stride*2, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\
+}\
+\
+static void OPNAME ## h264_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, int stride){\
+ uint8_t full[SIZE*(SIZE+5)];\
+ uint8_t * const full_mid= full + SIZE*2;\
+ int16_t tmp[SIZE*(SIZE+5)];\
+ uint8_t halfV[SIZE*SIZE];\
+ uint8_t halfHV[SIZE*SIZE];\
+ copy_block ## SIZE (full, src - stride*2 + 1, SIZE, stride, SIZE + 5);\
+ put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
+ put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
+ OPNAME ## pixels ## SIZE ## _l2(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\
+}\
+
+#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1)
+//#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
+#define op_put(a, b) a = cm[((b) + 16)>>5]
+#define op2_avg(a, b) a = (((a)+cm[((b) + 512)>>10]+1)>>1)
+#define op2_put(a, b) a = cm[((b) + 512)>>10]
+
+H264_LOWPASS(put_ , op_put, op2_put)
+H264_LOWPASS(avg_ , op_avg, op2_avg)
+H264_MC(put_, 4)
+H264_MC(put_, 8)
+H264_MC(put_, 16)
+H264_MC(avg_, 4)
+H264_MC(avg_, 8)
+H264_MC(avg_, 16)
+
+#undef op_avg
+#undef op_put
+#undef op2_avg
+#undef op2_put
+#endif
+
+#define op_scale1(x) block[x] = clip_uint8( (block[x]*weight + offset) >> log2_denom )
+#define op_scale2(x) dst[x] = clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
+#define H264_WEIGHT(W,H) \
+static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \
+ int attribute_unused x, y; \
+ offset <<= log2_denom; \
+ if(log2_denom) offset += 1<<(log2_denom-1); \
+ for(y=0; y<H; y++, block += stride){ \
+ op_scale1(0); \
+ op_scale1(1); \
+ if(W==2) continue; \
+ op_scale1(2); \
+ op_scale1(3); \
+ if(W==4) continue; \
+ op_scale1(4); \
+ op_scale1(5); \
+ op_scale1(6); \
+ op_scale1(7); \
+ if(W==8) continue; \
+ op_scale1(8); \
+ op_scale1(9); \
+ op_scale1(10); \
+ op_scale1(11); \
+ op_scale1(12); \
+ op_scale1(13); \
+ op_scale1(14); \
+ op_scale1(15); \
+ } \
+} \
+static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offsetd, int offsets){ \
+ int attribute_unused x, y; \
+ int offset = (offsets + offsetd + 1) >> 1; \
+ offset = ((offset << 1) + 1) << log2_denom; \
+ for(y=0; y<H; y++, dst += stride, src += stride){ \
+ op_scale2(0); \
+ op_scale2(1); \
+ if(W==2) continue; \
+ op_scale2(2); \
+ op_scale2(3); \
+ if(W==4) continue; \
+ op_scale2(4); \
+ op_scale2(5); \
+ op_scale2(6); \
+ op_scale2(7); \
+ if(W==8) continue; \
+ op_scale2(8); \
+ op_scale2(9); \
+ op_scale2(10); \
+ op_scale2(11); \
+ op_scale2(12); \
+ op_scale2(13); \
+ op_scale2(14); \
+ op_scale2(15); \
+ } \
+}
+
+H264_WEIGHT(16,16)
+H264_WEIGHT(16,8)
+H264_WEIGHT(8,16)
+H264_WEIGHT(8,8)
+H264_WEIGHT(8,4)
+H264_WEIGHT(4,8)
+H264_WEIGHT(4,4)
+H264_WEIGHT(4,2)
+H264_WEIGHT(2,4)
+H264_WEIGHT(2,2)
+
+#undef op_scale1
+#undef op_scale2
+#undef H264_WEIGHT
+
+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;
- src1+=srcStride;
- src2+=8;
- src3+=8;
- src4+=8;
+ 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 void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
+ int x;
+ const int strength= ff_h263_loop_filter_strength[qscale];
+
+ for(x=0; x<8; x++){
+ int d1, d2, ad1;
+ int p0= src[x-2*stride];
+ int p1= src[x-1*stride];
+ int p2= src[x+0*stride];
+ int p3= src[x+1*stride];
+ int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+ if (d<-2*strength) d1= 0;
+ else if(d<- strength) d1=-2*strength - d;
+ else if(d< strength) d1= d;
+ else if(d< 2*strength) d1= 2*strength - d;
+ else d1= 0;
+
+ p1 += d1;
+ p2 -= d1;
+ if(p1&256) p1= ~(p1>>31);
+ if(p2&256) p2= ~(p2>>31);
+
+ src[x-1*stride] = p1;
+ src[x+0*stride] = p2;
+
+ ad1= ABS(d1)>>1;
+
+ d2= clip((p0-p3)/4, -ad1, ad1);
+
+ src[x-2*stride] = p0 - d2;
+ src[x+ stride] = p3 + d2;
+ }
+}
+
+static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
+ int y;
+ const int strength= ff_h263_loop_filter_strength[qscale];
+
+ for(y=0; y<8; y++){
+ int d1, d2, ad1;
+ int p0= src[y*stride-2];
+ int p1= src[y*stride-1];
+ int p2= src[y*stride+0];
+ int p3= src[y*stride+1];
+ int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+ if (d<-2*strength) d1= 0;
+ else if(d<- strength) d1=-2*strength - d;
+ else if(d< strength) d1= d;
+ else if(d< 2*strength) d1= 2*strength - d;
+ else d1= 0;
+
+ p1 += d1;
+ p2 -= d1;
+ if(p1&256) p1= ~(p1>>31);
+ if(p2&256) p2= ~(p2>>31);
+
+ src[y*stride-1] = p1;
+ src[y*stride+0] = p2;
+
+ ad1= ABS(d1)>>1;
+
+ d2= clip((p0-p3)/4, -ad1, ad1);
+
+ src[y*stride-2] = p0 - d2;
+ src[y*stride+1] = p3 + d2;
}
}
-#define QPEL_MC(r, name) \
-static void qpel_mc00_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- put_block(dst, src, dstStride, srcStride);\
-}\
-\
-static void qpel_mc10_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 half[64];\
- qpel_h_lowpass(half, src, 8, srcStride, 8, 16-r);\
- avg2_block(dst, src, half, dstStride, srcStride, 1-r);\
-}\
-\
-static void qpel_mc20_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- qpel_h_lowpass(dst, src, dstStride, srcStride, 8, 16-r);\
-}\
-\
-static void qpel_mc30_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 half[64];\
- qpel_h_lowpass(half, src, 8, srcStride, 8, 16-r);\
- avg2_block(dst, src+1, half, dstStride, srcStride, 1-r);\
-}\
-\
-static void qpel_mc01_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 half[64];\
- qpel_v_lowpass(half, src, 8, srcStride, 8, 16-r);\
- avg2_block(dst, src, half, dstStride, srcStride, 1-r);\
-}\
-\
-static void qpel_mc02_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- qpel_v_lowpass(dst, src, dstStride, srcStride, 8, 16-r);\
-}\
-\
-static void qpel_mc03_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 half[64];\
- qpel_v_lowpass(half, src, 8, srcStride, 8, 16-r);\
- avg2_block(dst, src+srcStride, half, dstStride, srcStride, 1-r);\
-}\
-static void qpel_mc11_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg4_block(dst, src, halfH, halfV, halfHV, dstStride, srcStride, 2-r);\
-}\
-static void qpel_mc31_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src+1, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg4_block(dst, src+1, halfH, halfV, halfHV, dstStride, srcStride, 2-r);\
-}\
-static void qpel_mc13_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg4_block(dst, src+srcStride, halfH+8, halfV, halfHV, dstStride, srcStride, 2-r);\
-}\
-static void qpel_mc33_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src+1, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg4_block(dst, src+srcStride+1, halfH+8, halfV, halfHV, dstStride, srcStride, 2-r);\
-}\
-static void qpel_mc21_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg2_block(dst, halfH, halfHV, dstStride, 8, 1-r);\
-}\
-static void qpel_mc23_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg2_block(dst, halfH+8, halfHV, dstStride, 8, 1-r);\
-}\
-static void qpel_mc12_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg2_block(dst, halfV, halfHV, dstStride, 8, 1-r);\
-}\
-static void qpel_mc32_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(halfV, src+1, 8, srcStride, 8, 16-r);\
- qpel_v_lowpass(halfHV, halfH, 8, 8, 8, 16-r);\
- avg2_block(dst, halfV, halfHV, dstStride, 8, 1-r);\
-}\
-static void qpel_mc22_c ## name (UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int mx, int my)\
-{\
- UINT8 halfH[72];\
- qpel_h_lowpass(halfH, src, 8, srcStride, 9, 16-r);\
- qpel_v_lowpass(dst, halfH, dstStride, 8, 8, 16-r);\
-}\
-qpel_mc_func qpel_mc ## name ## _tab[16]={ \
- qpel_mc00_c ## name, \
- qpel_mc10_c ## name, \
- qpel_mc20_c ## name, \
- qpel_mc30_c ## name, \
- qpel_mc01_c ## name, \
- qpel_mc11_c ## name, \
- qpel_mc21_c ## name, \
- qpel_mc31_c ## name, \
- qpel_mc02_c ## name, \
- qpel_mc12_c ## name, \
- qpel_mc22_c ## name, \
- qpel_mc32_c ## name, \
- qpel_mc03_c ## name, \
- qpel_mc13_c ## name, \
- qpel_mc23_c ## name, \
- qpel_mc33_c ## name, \
-};
+static void h261_loop_filter_c(uint8_t *src, int stride){
+ int x,y,xy,yz;
+ int temp[64];
-QPEL_MC(0, _rnd)
-QPEL_MC(1, _no_rnd)
+ for(x=0; x<8; x++){
+ temp[x ] = 4*src[x ];
+ temp[x + 7*8] = 4*src[x + 7*stride];
+ }
+ for(y=1; y<7; y++){
+ for(x=0; x<8; x++){
+ xy = y * stride + x;
+ yz = y * 8 + x;
+ temp[yz] = src[xy - stride] + 2*src[xy] + src[xy + stride];
+ }
+ }
+
+ for(y=0; y<8; y++){
+ src[ y*stride] = (temp[ y*8] + 2)>>2;
+ src[7+y*stride] = (temp[7+y*8] + 2)>>2;
+ for(x=1; x<7; x++){
+ xy = y * stride + x;
+ yz = y * 8 + x;
+ src[xy] = (temp[yz-1] + 2*temp[yz] + temp[yz+1] + 8)>>4;
+ }
+ }
+}
-int pix_abs16x16_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - pix2[0]);
s += abs(pix1[1] - pix2[1]);
s += abs(pix1[2] - pix2[2]);
return s;
}
-int pix_abs16x16_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16_x2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
return s;
}
-int pix_abs16x16_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16_y2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
return s;
}
-int pix_abs16x16_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
return s;
}
-int pix_abs8x8_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static inline int pix_abs8_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - pix2[0]);
s += abs(pix1[1] - pix2[1]);
s += abs(pix1[2] - pix2[2]);
return s;
}
-int pix_abs8x8_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8_x2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
return s;
}
-int pix_abs8x8_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8_y2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
return s;
}
-int pix_abs8x8_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
return s;
}
-/* permute block according so that it corresponds to the MMX idct
- order */
-#ifdef SIMPLE_IDCT
- /* general permutation, but perhaps slightly slower */
-void block_permute(INT16 *block)
+static int nsse16_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score1=0;
+ int score2=0;
+ int x,y;
+
+ for(y=0; y<h; y++){
+ for(x=0; x<16; x++){
+ score1+= (s1[x ] - s2[x ])*(s1[x ] - s2[x ]);
+ }
+ if(y+1<h){
+ for(x=0; x<15; x++){
+ score2+= ABS( s1[x ] - s1[x +stride]
+ - s1[x+1] + s1[x+1+stride])
+ -ABS( s2[x ] - s2[x +stride]
+ - s2[x+1] + s2[x+1+stride]);
+ }
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
+ else return score1 + ABS(score2)*8;
+}
+
+static int nsse8_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score1=0;
+ int score2=0;
+ int x,y;
+
+ for(y=0; y<h; y++){
+ for(x=0; x<8; x++){
+ score1+= (s1[x ] - s2[x ])*(s1[x ] - s2[x ]);
+ }
+ if(y+1<h){
+ for(x=0; x<7; x++){
+ score2+= ABS( s1[x ] - s1[x +stride]
+ - s1[x+1] + s1[x+1+stride])
+ -ABS( s2[x ] - s2[x +stride]
+ - s2[x+1] + s2[x+1+stride]);
+ }
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
+ else return score1 + ABS(score2)*8;
+}
+
+static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale){
+ int i;
+ unsigned int sum=0;
+
+ for(i=0; i<8*8; i++){
+ int b= rem[i] + ((basis[i]*scale + (1<<(BASIS_SHIFT - RECON_SHIFT-1)))>>(BASIS_SHIFT - RECON_SHIFT));
+ int w= weight[i];
+ b>>= RECON_SHIFT;
+ assert(-512<b && b<512);
+
+ sum += (w*b)*(w*b)>>4;
+ }
+ return sum>>2;
+}
+
+static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){
+ int i;
+
+ for(i=0; i<8*8; i++){
+ rem[i] += (basis[i]*scale + (1<<(BASIS_SHIFT - RECON_SHIFT-1)))>>(BASIS_SHIFT - RECON_SHIFT);
+ }
+}
+
+/**
+ * permutes an 8x8 block.
+ * @param block the block which will be permuted according to the given permutation vector
+ * @param permutation the permutation vector
+ * @param last the last non zero coefficient in scantable order, used to speed the permutation up
+ * @param scantable the used scantable, this is only used to speed the permutation up, the block is not
+ * (inverse) permutated to scantable order!
+ */
+void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last)
{
- int i;
- INT16 temp[64];
+ 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<64; i++) temp[ block_permute_op(i) ] = block[i];
+ 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];
+ }
+}
- for(i=0; i<64; i++) block[i] = temp[i];
+static int zero_cmp(void *s, uint8_t *a, uint8_t *b, int stride, int h){
+ return 0;
+}
+
+void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
+ int i;
+
+ memset(cmp, 0, sizeof(void*)*5);
+
+ for(i=0; i<5; i++){
+ switch(type&0xFF){
+ case FF_CMP_SAD:
+ cmp[i]= c->sad[i];
+ break;
+ case FF_CMP_SATD:
+ cmp[i]= c->hadamard8_diff[i];
+ break;
+ case FF_CMP_SSE:
+ cmp[i]= c->sse[i];
+ break;
+ case FF_CMP_DCT:
+ cmp[i]= c->dct_sad[i];
+ break;
+ case FF_CMP_DCTMAX:
+ cmp[i]= c->dct_max[i];
+ break;
+ case FF_CMP_PSNR:
+ cmp[i]= c->quant_psnr[i];
+ break;
+ case FF_CMP_BIT:
+ cmp[i]= c->bit[i];
+ break;
+ case FF_CMP_RD:
+ cmp[i]= c->rd[i];
+ break;
+ case FF_CMP_VSAD:
+ cmp[i]= c->vsad[i];
+ break;
+ case FF_CMP_VSSE:
+ cmp[i]= c->vsse[i];
+ break;
+ case FF_CMP_ZERO:
+ cmp[i]= zero_cmp;
+ break;
+ case FF_CMP_NSSE:
+ cmp[i]= c->nsse[i];
+ break;
+ case FF_CMP_W53:
+ cmp[i]= c->w53[i];
+ break;
+ case FF_CMP_W97:
+ cmp[i]= c->w97[i];
+ break;
+ default:
+ av_log(NULL, AV_LOG_ERROR,"internal error in cmp function selection\n");
+ }
+ }
}
-#else
-void block_permute(INT16 *block)
+/**
+ * memset(blocks, 0, sizeof(DCTELEM)*6*64)
+ */
+static void clear_blocks_c(DCTELEM *blocks)
{
- int tmp1, tmp2, tmp3, tmp4, tmp5, tmp6;
+ memset(blocks, 0, sizeof(DCTELEM)*6*64);
+}
+
+static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){
+ int i;
+ for(i=0; i+7<w; i+=8){
+ 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];
+}
+
+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+=8){
+ 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<8;i++) {
- tmp1 = block[1];
- tmp2 = block[2];
- tmp3 = block[3];
- tmp4 = block[4];
- tmp5 = block[5];
- tmp6 = block[6];
- block[1] = tmp2;
- block[2] = tmp4;
- block[3] = tmp6;
- block[4] = tmp1;
- block[5] = tmp3;
- block[6] = tmp5;
- block += 8;
+static void sub_hfyu_median_prediction_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top){
+ int i;
+ uint8_t l, lt;
+
+ l= *left;
+ lt= *left_top;
+
+ for(i=0; i<w; i++){
+ const int pred= mid_pred(l, src1[i], (l + src1[i] - lt)&0xFF);
+ lt= src1[i];
+ l= src2[i];
+ dst[i]= l - pred;
+ }
+
+ *left= l;
+ *left_top= lt;
+}
+
+#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_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
+ int i;
+ int temp[64];
+ int sum=0;
+
+ assert(h==8);
+
+ 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_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_t *dummy, int stride, int h){
+ int i;
+ int temp[64];
+ int sum=0;
+
+ assert(h==8);
+
+ for(i=0; i<8; i++){
+ //FIXME try pointer walks
+ BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0],src[stride*i+1]);
+ BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2],src[stride*i+3]);
+ BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4],src[stride*i+5]);
+ BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6],src[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]);
+ }
+
+ sum -= ABS(temp[8*0] + temp[8*4]); // -mean
+
+ return sum;
+}
+
+static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
+ int sum=0, i;
+
+ assert(h==8);
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+ s->dsp.fdct(temp);
+
+ for(i=0; i<64; i++)
+ sum+= ABS(temp[i]);
+
+ return sum;
+}
-void clear_blocks_c(DCTELEM *blocks)
+static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
+ int sum=0, i;
+
+ assert(h==8);
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+ s->dsp.fdct(temp);
+
+ for(i=0; i<64; i++)
+ sum= FFMAX(sum, ABS(temp[i]));
+
+ return sum;
+}
+
+void simple_idct(DCTELEM *block); //FIXME
+
+static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64*2/8];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
+ DCTELEM * const bak = ((DCTELEM*)aligned_temp)+64;
+ int sum=0, i;
+
+ assert(h==8);
+ s->mb_intra=0;
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+
+ memcpy(bak, temp, 64*sizeof(DCTELEM));
+
+ s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
+ s->dct_unquantize_inter(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;
+}
+
+static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ const uint8_t *scantable= s->intra_scantable.permutated;
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+ uint64_t __align8 aligned_bak[stride];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
+ uint8_t * const bak= (uint8_t*)aligned_bak;
+ int i, last, run, bits, level, distoration, start_i;
+ const int esc_length= s->ac_esc_length;
+ uint8_t * length;
+ uint8_t * last_length;
+
+ assert(h==8);
+
+ for(i=0; i<8; i++){
+ ((uint32_t*)(bak + i*stride))[0]= ((uint32_t*)(src2 + i*stride))[0];
+ ((uint32_t*)(bak + i*stride))[1]= ((uint32_t*)(src2 + i*stride))[1];
+ }
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+
+ s->block_last_index[0/*FIXME*/]= last= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
+
+ bits=0;
+
+ if (s->mb_intra) {
+ start_i = 1;
+ length = s->intra_ac_vlc_length;
+ last_length= s->intra_ac_vlc_last_length;
+ bits+= s->luma_dc_vlc_length[temp[0] + 256]; //FIXME chroma
+ } else {
+ start_i = 0;
+ length = s->inter_ac_vlc_length;
+ last_length= s->inter_ac_vlc_last_length;
+ }
+
+ if(last>=start_i){
+ run=0;
+ for(i=start_i; i<last; i++){
+ int j= scantable[i];
+ level= temp[j];
+
+ if(level){
+ level+=64;
+ if((level&(~127)) == 0){
+ bits+= length[UNI_AC_ENC_INDEX(run, level)];
+ }else
+ bits+= esc_length;
+ run=0;
+ }else
+ run++;
+ }
+ i= scantable[last];
+
+ level= temp[i] + 64;
+
+ assert(level - 64);
+
+ if((level&(~127)) == 0){
+ bits+= last_length[UNI_AC_ENC_INDEX(run, level)];
+ }else
+ bits+= esc_length;
+
+ }
+
+ if(last>=0){
+ if(s->mb_intra)
+ s->dct_unquantize_intra(s, temp, 0, s->qscale);
+ else
+ s->dct_unquantize_inter(s, temp, 0, s->qscale);
+ }
+
+ s->dsp.idct_add(bak, stride, temp);
+
+ distoration= s->dsp.sse[1](NULL, bak, src1, stride, 8);
+
+ return distoration + ((bits*s->qscale*s->qscale*109 + 64)>>7);
+}
+
+static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+ MpegEncContext * const s= (MpegEncContext *)c;
+ const uint8_t *scantable= s->intra_scantable.permutated;
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
+ int i, last, run, bits, level, start_i;
+ const int esc_length= s->ac_esc_length;
+ uint8_t * length;
+ uint8_t * last_length;
+
+ assert(h==8);
+
+ s->dsp.diff_pixels(temp, src1, src2, stride);
+
+ s->block_last_index[0/*FIXME*/]= last= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
+
+ bits=0;
+
+ if (s->mb_intra) {
+ start_i = 1;
+ length = s->intra_ac_vlc_length;
+ last_length= s->intra_ac_vlc_last_length;
+ bits+= s->luma_dc_vlc_length[temp[0] + 256]; //FIXME chroma
+ } else {
+ start_i = 0;
+ length = s->inter_ac_vlc_length;
+ last_length= s->inter_ac_vlc_last_length;
+ }
+
+ if(last>=start_i){
+ run=0;
+ for(i=start_i; i<last; i++){
+ int j= scantable[i];
+ level= temp[j];
+
+ if(level){
+ level+=64;
+ if((level&(~127)) == 0){
+ bits+= length[UNI_AC_ENC_INDEX(run, level)];
+ }else
+ bits+= esc_length;
+ run=0;
+ }else
+ run++;
+ }
+ i= scantable[last];
+
+ level= temp[i] + 64;
+
+ assert(level - 64);
+
+ if((level&(~127)) == 0){
+ bits+= last_length[UNI_AC_ENC_INDEX(run, level)];
+ }else
+ bits+= esc_length;
+ }
+
+ return bits;
+}
+
+static int vsad_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x+=4){
+ score+= ABS(s[x ] - s[x +stride]) + ABS(s[x+1] - s[x+1+stride])
+ +ABS(s[x+2] - s[x+2+stride]) + ABS(s[x+3] - s[x+3+stride]);
+ }
+ s+= stride;
+ }
+
+ return score;
+}
+
+static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x++){
+ score+= ABS(s1[x ] - s2[x ] - s1[x +stride] + s2[x +stride]);
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ return score;
+}
+
+#define SQ(a) ((a)*(a))
+static int vsse_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x+=4){
+ score+= SQ(s[x ] - s[x +stride]) + SQ(s[x+1] - s[x+1+stride])
+ +SQ(s[x+2] - s[x+2+stride]) + SQ(s[x+3] - s[x+3+stride]);
+ }
+ s+= stride;
+ }
+
+ return score;
+}
+
+static int vsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x++){
+ score+= SQ(s1[x ] - s2[x ] - s1[x +stride] + s2[x +stride]);
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ return score;
+}
+
+WARPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
+WARPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
+WARPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
+WARPER8_16_SQ(dct_max8x8_c, dct_max16_c)
+WARPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
+WARPER8_16_SQ(rd8x8_c, rd16_c)
+WARPER8_16_SQ(bit8x8_c, bit16_c)
+
+/* XXX: those functions should be suppressed ASAP when all IDCTs are
+ converted */
+static void ff_jref_idct_put(uint8_t *dest, int line_size, DCTELEM *block)
{
- memset(blocks, 0, sizeof(DCTELEM)*6*64);
+ j_rev_dct (block);
+ put_pixels_clamped_c(block, dest, line_size);
+}
+static void ff_jref_idct_add(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ j_rev_dct (block);
+ add_pixels_clamped_c(block, dest, line_size);
}
-void dsputil_init(void)
+static void ff_jref_idct4_put(uint8_t *dest, int line_size, DCTELEM *block)
{
- int i, j;
- int use_permuted_idct;
+ j_rev_dct4 (block);
+ put_pixels_clamped4_c(block, dest, line_size);
+}
+static void ff_jref_idct4_add(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ j_rev_dct4 (block);
+ add_pixels_clamped4_c(block, dest, line_size);
+}
+
+static void ff_jref_idct2_put(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ j_rev_dct2 (block);
+ put_pixels_clamped2_c(block, dest, line_size);
+}
+static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ j_rev_dct2 (block);
+ add_pixels_clamped2_c(block, dest, line_size);
+}
+
+static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ dest[0] = cm[(block[0] + 4)>>3];
+}
+static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block)
+{
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
+
+ dest[0] = cm[dest[0] + ((block[0] + 4)>>3)];
+}
+
+/* init static data */
+void dsputil_static_init(void)
+{
+ int i;
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;
+}
-#ifdef SIMPLE_IDCT
- ff_idct = simple_idct;
-#else
- ff_idct = j_rev_dct;
-#endif
- 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_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;
- av_fdct = jpeg_fdct_ifast;
-
- use_permuted_idct = 1;
+
+void dsputil_init(DSPContext* c, AVCodecContext *avctx)
+{
+ int i;
+
+#ifdef CONFIG_ENCODERS
+ if(avctx->dct_algo==FF_DCT_FASTINT) {
+ c->fdct = fdct_ifast;
+ c->fdct248 = fdct_ifast248;
+ }
+ else if(avctx->dct_algo==FF_DCT_FAAN) {
+ c->fdct = ff_faandct;
+ c->fdct248 = ff_faandct248;
+ }
+ else {
+ c->fdct = ff_jpeg_fdct_islow; //slow/accurate/default
+ c->fdct248 = ff_fdct248_islow;
+ }
+#endif //CONFIG_ENCODERS
+
+ if(avctx->lowres==1){
+ if(avctx->idct_algo==FF_IDCT_INT || avctx->idct_algo==FF_IDCT_AUTO){
+ c->idct_put= ff_jref_idct4_put;
+ c->idct_add= ff_jref_idct4_add;
+ }else{
+ c->idct_put= ff_h264_lowres_idct_put_c;
+ c->idct_add= ff_h264_lowres_idct_add_c;
+ }
+ c->idct = j_rev_dct4;
+ c->idct_permutation_type= FF_NO_IDCT_PERM;
+ }else if(avctx->lowres==2){
+ c->idct_put= ff_jref_idct2_put;
+ c->idct_add= ff_jref_idct2_add;
+ c->idct = j_rev_dct2;
+ c->idct_permutation_type= FF_NO_IDCT_PERM;
+ }else if(avctx->lowres==3){
+ c->idct_put= ff_jref_idct1_put;
+ c->idct_add= ff_jref_idct1_add;
+ c->idct = j_rev_dct1;
+ c->idct_permutation_type= FF_NO_IDCT_PERM;
+ }else{
+ if(avctx->idct_algo==FF_IDCT_INT){
+ c->idct_put= ff_jref_idct_put;
+ c->idct_add= ff_jref_idct_add;
+ c->idct = j_rev_dct;
+ c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
+ }else{ //accurate/default
+ c->idct_put= simple_idct_put;
+ c->idct_add= simple_idct_add;
+ c->idct = simple_idct;
+ c->idct_permutation_type= FF_NO_IDCT_PERM;
+ }
+ }
+
+ c->h264_idct_add= ff_h264_idct_add_c;
+
+ /* VP3 DSP support */
+ c->vp3_dsp_init = vp3_dsp_init_c;
+ c->vp3_idct = vp3_idct_c;
+
+ c->get_pixels = get_pixels_c;
+ c->diff_pixels = diff_pixels_c;
+ c->put_pixels_clamped = put_pixels_clamped_c;
+ c->put_signed_pixels_clamped = put_signed_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;
+
+ /* TODO [0] 16 [1] 8 */
+ c->pix_abs[0][0] = pix_abs16_c;
+ c->pix_abs[0][1] = pix_abs16_x2_c;
+ c->pix_abs[0][2] = pix_abs16_y2_c;
+ c->pix_abs[0][3] = pix_abs16_xy2_c;
+ c->pix_abs[1][0] = pix_abs8_c;
+ c->pix_abs[1][1] = pix_abs8_x2_c;
+ c->pix_abs[1][2] = pix_abs8_y2_c;
+ c->pix_abs[1][3] = pix_abs8_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(put, 2, 4);
+ dspfunc(put, 3, 2);
+
+ dspfunc(avg, 0, 16);
+ dspfunc(avg_no_rnd, 0, 16);
+ dspfunc(avg, 1, 8);
+ dspfunc(avg_no_rnd, 1, 8);
+ dspfunc(avg, 2, 4);
+ dspfunc(avg, 3, 2);
+#undef dspfunc
+
+ c->put_no_rnd_pixels_l2[0]= put_no_rnd_pixels16_l2_c;
+ c->put_no_rnd_pixels_l2[1]= put_no_rnd_pixels8_l2_c;
+
+ c->put_tpel_pixels_tab[ 0] = put_tpel_pixels_mc00_c;
+ c->put_tpel_pixels_tab[ 1] = put_tpel_pixels_mc10_c;
+ c->put_tpel_pixels_tab[ 2] = put_tpel_pixels_mc20_c;
+ c->put_tpel_pixels_tab[ 4] = put_tpel_pixels_mc01_c;
+ c->put_tpel_pixels_tab[ 5] = put_tpel_pixels_mc11_c;
+ c->put_tpel_pixels_tab[ 6] = put_tpel_pixels_mc21_c;
+ c->put_tpel_pixels_tab[ 8] = put_tpel_pixels_mc02_c;
+ c->put_tpel_pixels_tab[ 9] = put_tpel_pixels_mc12_c;
+ c->put_tpel_pixels_tab[10] = put_tpel_pixels_mc22_c;
+
+ c->avg_tpel_pixels_tab[ 0] = avg_tpel_pixels_mc00_c;
+ c->avg_tpel_pixels_tab[ 1] = avg_tpel_pixels_mc10_c;
+ c->avg_tpel_pixels_tab[ 2] = avg_tpel_pixels_mc20_c;
+ c->avg_tpel_pixels_tab[ 4] = avg_tpel_pixels_mc01_c;
+ c->avg_tpel_pixels_tab[ 5] = avg_tpel_pixels_mc11_c;
+ c->avg_tpel_pixels_tab[ 6] = avg_tpel_pixels_mc21_c;
+ c->avg_tpel_pixels_tab[ 8] = avg_tpel_pixels_mc02_c;
+ c->avg_tpel_pixels_tab[ 9] = avg_tpel_pixels_mc12_c;
+ c->avg_tpel_pixels_tab[10] = avg_tpel_pixels_mc22_c;
+
+#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); */
+
+ dspfunc(put_h264_qpel, 0, 16);
+ dspfunc(put_h264_qpel, 1, 8);
+ dspfunc(put_h264_qpel, 2, 4);
+ dspfunc(avg_h264_qpel, 0, 16);
+ dspfunc(avg_h264_qpel, 1, 8);
+ dspfunc(avg_h264_qpel, 2, 4);
+
+#undef dspfunc
+ c->put_h264_chroma_pixels_tab[0]= put_h264_chroma_mc8_c;
+ c->put_h264_chroma_pixels_tab[1]= put_h264_chroma_mc4_c;
+ c->put_h264_chroma_pixels_tab[2]= put_h264_chroma_mc2_c;
+ c->avg_h264_chroma_pixels_tab[0]= avg_h264_chroma_mc8_c;
+ c->avg_h264_chroma_pixels_tab[1]= avg_h264_chroma_mc4_c;
+ c->avg_h264_chroma_pixels_tab[2]= avg_h264_chroma_mc2_c;
+
+ c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c;
+ c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c;
+ c->weight_h264_pixels_tab[2]= weight_h264_pixels8x16_c;
+ c->weight_h264_pixels_tab[3]= weight_h264_pixels8x8_c;
+ c->weight_h264_pixels_tab[4]= weight_h264_pixels8x4_c;
+ c->weight_h264_pixels_tab[5]= weight_h264_pixels4x8_c;
+ c->weight_h264_pixels_tab[6]= weight_h264_pixels4x4_c;
+ c->weight_h264_pixels_tab[7]= weight_h264_pixels4x2_c;
+ c->weight_h264_pixels_tab[8]= weight_h264_pixels2x4_c;
+ c->weight_h264_pixels_tab[9]= weight_h264_pixels2x2_c;
+ c->biweight_h264_pixels_tab[0]= biweight_h264_pixels16x16_c;
+ c->biweight_h264_pixels_tab[1]= biweight_h264_pixels16x8_c;
+ c->biweight_h264_pixels_tab[2]= biweight_h264_pixels8x16_c;
+ c->biweight_h264_pixels_tab[3]= biweight_h264_pixels8x8_c;
+ c->biweight_h264_pixels_tab[4]= biweight_h264_pixels8x4_c;
+ c->biweight_h264_pixels_tab[5]= biweight_h264_pixels4x8_c;
+ c->biweight_h264_pixels_tab[6]= biweight_h264_pixels4x4_c;
+ c->biweight_h264_pixels_tab[7]= biweight_h264_pixels4x2_c;
+ c->biweight_h264_pixels_tab[8]= biweight_h264_pixels2x4_c;
+ c->biweight_h264_pixels_tab[9]= biweight_h264_pixels2x2_c;
+
+ 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;
+
+#define SET_CMP_FUNC(name) \
+ c->name[0]= name ## 16_c;\
+ c->name[1]= name ## 8x8_c;
+
+ SET_CMP_FUNC(hadamard8_diff)
+ c->hadamard8_diff[4]= hadamard8_intra16_c;
+ SET_CMP_FUNC(dct_sad)
+ SET_CMP_FUNC(dct_max)
+ c->sad[0]= pix_abs16_c;
+ c->sad[1]= pix_abs8_c;
+ c->sse[0]= sse16_c;
+ c->sse[1]= sse8_c;
+ c->sse[2]= sse4_c;
+ SET_CMP_FUNC(quant_psnr)
+ SET_CMP_FUNC(rd)
+ SET_CMP_FUNC(bit)
+ c->vsad[0]= vsad16_c;
+ c->vsad[4]= vsad_intra16_c;
+ c->vsse[0]= vsse16_c;
+ c->vsse[4]= vsse_intra16_c;
+ c->nsse[0]= nsse16_c;
+ c->nsse[1]= nsse8_c;
+ c->w53[0]= w53_16_c;
+ c->w53[1]= w53_8_c;
+ c->w97[0]= w97_16_c;
+ c->w97[1]= w97_8_c;
+
+ c->add_bytes= add_bytes_c;
+ c->diff_bytes= diff_bytes_c;
+ c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_c;
+ c->bswap_buf= bswap_buf;
+
+ c->h263_h_loop_filter= h263_h_loop_filter_c;
+ c->h263_v_loop_filter= h263_v_loop_filter_c;
+
+ c->h261_loop_filter= h261_loop_filter_c;
+
+ c->try_8x8basis= try_8x8basis_c;
+ c->add_8x8basis= add_8x8basis_c;
#ifdef HAVE_MMX
- dsputil_init_mmx();
+ dsputil_init_mmx(c, avctx);
#endif
#ifdef ARCH_ARMV4L
- dsputil_init_armv4l();
+ dsputil_init_armv4l(c, avctx);
#endif
#ifdef HAVE_MLIB
- dsputil_init_mlib();
- use_permuted_idct = 0;
+ dsputil_init_mlib(c, avctx);
+#endif
+#ifdef ARCH_SPARC
+ dsputil_init_vis(c,avctx);
#endif
#ifdef ARCH_ALPHA
- dsputil_init_alpha();
- use_permuted_idct = 0;
+ dsputil_init_alpha(c, avctx);
#endif
-
-#ifdef SIMPLE_IDCT
- if(ff_idct == simple_idct) use_permuted_idct=0;
+#ifdef ARCH_POWERPC
+ dsputil_init_ppc(c, avctx);
#endif
-
- if(use_permuted_idct)
-#ifdef SIMPLE_IDCT
- for(i=0; i<64; i++) permutation[i]= simple_mmx_permutation[i];
-#else
- for(i=0; i<64; i++) permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2);
+#ifdef HAVE_MMI
+ dsputil_init_mmi(c, avctx);
#endif
- else
- for(i=0; i<64; i++) permutation[i]=i;
-
- for(i=0; i<64; i++) inv_zigzag_direct16[zigzag_direct[i]]= i+1;
- for(i=0; i<64; i++) zigzag_direct_noperm[i]= zigzag_direct[i];
-
- if (use_permuted_idct) {
- /* permute for IDCT */
- for(i=0;i<64;i++) {
- j = zigzag_direct[i];
- zigzag_direct[i] = block_permute_op(j);
- j = ff_alternate_horizontal_scan[i];
- ff_alternate_horizontal_scan[i] = block_permute_op(j);
- j = ff_alternate_vertical_scan[i];
- ff_alternate_vertical_scan[i] = block_permute_op(j);
- }
- block_permute(default_intra_matrix);
- block_permute(default_non_intra_matrix);
- block_permute(ff_mpeg4_default_intra_matrix);
- block_permute(ff_mpeg4_default_non_intra_matrix);
- }
-
- build_zigzag_end();
-}
-
-/* remove any non bit exact operation (testing purpose) */
-void avcodec_set_bit_exact(void)
-{
-#ifdef HAVE_MMX
- dsputil_set_bit_exact_mmx();
+#ifdef ARCH_SH4
+ dsputil_init_sh4(c,avctx);
#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;
+ switch(c->idct_permutation_type){
+ case FF_NO_IDCT_PERM:
+ for(i=0; i<64; i++)
+ c->idct_permutation[i]= i;
+ break;
+ case FF_LIBMPEG2_IDCT_PERM:
+ for(i=0; i<64; i++)
+ c->idct_permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2);
+ break;
+ case FF_SIMPLE_IDCT_PERM:
+ for(i=0; i<64; i++)
+ c->idct_permutation[i]= simple_mmx_permutation[i];
+ break;
+ case FF_TRANSPOSE_IDCT_PERM:
+ for(i=0; i<64; i++)
+ c->idct_permutation[i]= ((i&7)<<3) | (i>>3);
+ break;
+ default:
+ av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n");
}
-
- 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;
}
projects / ffmpeg / blobdiff