*
* 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"
-int ff_bit_exact=0;
-UINT8 cropTbl[256 + 2 * MAX_NEG_CROP];
-UINT32 squareTbl[512];
+uint8_t cropTbl[256 + 2 * MAX_NEG_CROP];
+uint32_t squareTbl[512];
-const UINT8 ff_zigzag_direct[64] = {
+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,
};
/* not permutated inverse zigzag_direct + 1 for MMX quantizer */
-UINT16 __align8 inv_zigzag_direct16[64];
+uint16_t __align8 inv_zigzag_direct16[64];
-const UINT8 ff_alternate_horizontal_scan[64] = {
+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,
52, 53, 54, 55, 60, 61, 62, 63,
};
-const UINT8 ff_alternate_vertical_scan[64] = {
+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,
};
/* a*inverse[b]>>32 == a/b for all 0<=a<=65536 && 2<=b<=255 */
-const 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,
};
-static int pix_sum_c(UINT8 * pix, int line_size)
+/* 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;
return s;
}
-static int pix_norm1_c(UINT8 * pix, int line_size)
+static int pix_norm1_c(uint8_t * pix, int line_size)
{
int s, i, j;
- UINT32 *sq = squareTbl + 256;
+ 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[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;
}
+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 int sse8_c(void *v, UINT8 * pix1, UINT8 * pix2, int line_size)
+static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size)
{
int s, i;
- UINT32 *sq = squareTbl + 256;
+ uint32_t *sq = squareTbl + 256;
s = 0;
for (i = 0; i < 8; i++) {
return s;
}
-static int sse16_c(void *v, UINT8 * pix1, UINT8 * pix2, int line_size)
+static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
{
- int s, i, j;
- UINT32 *sq = squareTbl + 256;
+ int s, i;
+ uint32_t *sq = squareTbl + 256;
s = 0;
for (i = 0; i < 16; i++) {
- for (j = 0; j < 16; j += 8) {
- s += sq[pix1[0] - pix2[0]];
- s += sq[pix1[1] - pix2[1]];
- s += sq[pix1[2] - pix2[2]];
- s += sq[pix1[3] - pix2[3]];
- s += sq[pix1[4] - pix2[4]];
- s += sq[pix1[5] - pix2[5]];
- s += sq[pix1[6] - pix2[6]];
- s += sq[pix1[7] - pix2[7]];
- pix1 += 8;
- pix2 += 8;
- }
- pix1 += line_size - 16;
- pix2 += line_size - 16;
+ 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 void get_pixels_c(DCTELEM *restrict block, const UINT8 *pixels, int line_size)
+static void get_pixels_c(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
{
int i;
}
}
-static void diff_pixels_c(DCTELEM *restrict block, const UINT8 *s1,
- const UINT8 *s2, int stride){
+static void diff_pixels_c(DCTELEM *restrict block, const uint8_t *s1,
+ const uint8_t *s2, int stride){
int i;
/* read the pixels */
}
-static void put_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
+static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
int line_size)
{
int i;
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
/* read the pixels */
for(i=0;i<8;i++) {
}
}
-static void add_pixels_clamped_c(const DCTELEM *block, UINT8 *restrict pixels,
+static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
int line_size)
{
int i;
- UINT8 *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
/* read the pixels */
for(i=0;i<8;i++) {
#else // 64 bit variant
#define PIXOP2(OPNAME, OP) \
+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++){\
uint32_t a,b;\
a= LD32(&src1[i*src_stride1 ]);\
b= LD32(&src2[i*src_stride2 ]);\
- OP(*((uint32_t*)&dst[i*dst_stride ]), (a&b) + (((a^b)&0xFEFEFEFEUL)>>1));\
+ 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]), (a&b) + (((a^b)&0xFEFEFEFEUL)>>1));\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), no_rnd_avg32(a, b));\
}\
}\
\
uint32_t a,b;\
a= LD32(&src1[i*src_stride1 ]);\
b= LD32(&src2[i*src_stride2 ]);\
- OP(*((uint32_t*)&dst[i*dst_stride ]), (a|b) - (((a^b)&0xFEFEFEFEUL)>>1));\
+ 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]), (a|b) - (((a^b)&0xFEFEFEFEUL)>>1));\
+ OP(*((uint32_t*)&dst[i*dst_stride+4]), rnd_avg32(a, b));\
+ }\
+}\
+\
+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++){\
+ 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));\
+ }\
+}\
+\
+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++){\
+ 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));\
}\
}\
\
OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
}\
}\
+\
+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;\
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 i;\
+ const uint32_t a= LD32(pixels );\
+ const uint32_t b= LD32(pixels+1);\
+ uint32_t l0= (a&0x03030303UL)\
+ + (b&0x03030303UL)\
+ + 0x02020202UL;\
+ uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ uint32_t l1,h1;\
+\
+ 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)\
+ + 0x02020202UL;\
+ h0= ((a&0xFCFCFCFCUL)>>2)\
+ + ((b&0xFCFCFCFCUL)>>2);\
+ OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
+ pixels+=line_size;\
+ block +=line_size;\
+ }\
+}\
+\
static inline void OPNAME ## _pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
int j;\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels8_y2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels8_xy2_c, 8)\
-#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEUL)>>1) )
+#define op_avg(a, b) a = rnd_avg32(a, b)
#endif
#define op_put(a, b) a = b
#define avg4(a,b,c,d) ((a+b+c+d+2)>>2)
-static void gmc1_c(UINT8 *dst, UINT8 *src, int stride, int h, int x16, int y16, int rounder)
+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);
}
}
-static void gmc_c(UINT8 *dst, UINT8 *src, int stride, int h, int ox, int oy,
+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;
}
}
-static inline void copy_block17(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h)
+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
+
+#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)
+{
+ int i;
+ for(i=0; i<h; i++)
+ {
+ ST32(dst , LD32(src ));
+ dst+=dstStride;
+ src+=srcStride;
+ }
+}
+
+static inline void copy_block8(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
+{
+ int i;
+ for(i=0; i<h; i++)
+ {
+ ST32(dst , LD32(src ));
+ ST32(dst+4 , LD32(src+4 ));
+ dst+=dstStride;
+ src+=srcStride;
+ }
+}
+
+static inline void copy_block16(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
+{
+ int i;
+ for(i=0; i<h; i++)
+ {
+ 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 copy_block17(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
int i;
for(i=0; i<h; i++)
}
}
-static inline void copy_block9(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h)
+static inline void copy_block9(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
int i;
for(i=0; i<h; i++)
#define QPEL_MC(r, OPNAME, RND, OP) \
-static void OPNAME ## mpeg4_qpel8_h_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h){\
- UINT8 *cm = cropTbl + MAX_NEG_CROP;\
+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++)\
{\
}\
}\
\
-static void OPNAME ## mpeg4_qpel8_v_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride){\
+static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
const int w=8;\
- UINT8 *cm = cropTbl + MAX_NEG_CROP;\
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;\
int i;\
for(i=0; i<w; i++)\
{\
}\
}\
\
-static void OPNAME ## mpeg4_qpel16_h_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride, int h){\
- UINT8 *cm = cropTbl + MAX_NEG_CROP;\
+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++)\
}\
}\
\
-static void OPNAME ## mpeg4_qpel16_v_lowpass(UINT8 *dst, UINT8 *src, int dstStride, int srcStride){\
- UINT8 *cm = cropTbl + MAX_NEG_CROP;\
+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++)\
}\
}\
\
-static void OPNAME ## qpel8_mc00_c (UINT8 *dst, UINT8 *src, int stride){\
+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 *dst, UINT8 *src, int stride){\
- UINT8 half[64];\
+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 *dst, UINT8 *src, int stride){\
+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 *dst, UINT8 *src, int stride){\
- UINT8 half[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 half[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 half[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[72];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
- UINT8 halfV[64];\
- UINT8 halfHV[64];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[16*9];\
- UINT8 halfH[72];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[72];\
+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 *dst, UINT8 *src, int stride){\
+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 *dst, UINT8 *src, int stride){\
- UINT8 half[256];\
+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 *dst, UINT8 *src, int stride){\
+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 *dst, UINT8 *src, int stride){\
- UINT8 half[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 half[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 half[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[272];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
- UINT8 halfV[256];\
- UINT8 halfHV[256];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 full[24*17];\
- UINT8 halfH[272];\
+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 *dst, UINT8 *src, int stride){\
- UINT8 halfH[272];\
+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);\
}
#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;\
+ tmp += 8*tmpStride;\
+ 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
+
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;
}
-static inline int pix_abs16x16_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static inline int pix_abs16x16_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
return s;
}
-static int pix_abs16x16_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
return s;
}
-static int pix_abs16x16_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
for(i=0;i<16;i++) {
return s;
}
-static int pix_abs16x16_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs16x16_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
for(i=0;i<16;i++) {
return s;
}
-static inline int pix_abs8x8_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static inline int pix_abs8x8_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
return s;
}
-static int pix_abs8x8_x2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
return s;
}
-static int pix_abs8x8_y2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
for(i=0;i<8;i++) {
return s;
}
-static int pix_abs8x8_xy2_c(UINT8 *pix1, UINT8 *pix2, int line_size)
+static int pix_abs8x8_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
{
int s, i;
- UINT8 *pix3 = pix2 + line_size;
+ uint8_t *pix3 = pix2 + line_size;
s = 0;
for(i=0;i<8;i++) {
return pix_abs8x8_c(a,b,stride);
}
-void ff_block_permute(DCTELEM *block, UINT8 *permutation, const UINT8 *scantable, int last)
+/**
+ * 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;
DCTELEM temp[64];
}
}
+/**
+ * memset(blocks, 0, sizeof(DCTELEM)*6*64)
+ */
static void clear_blocks_c(DCTELEM *blocks)
{
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++){
+ 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];
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++){
+ 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];
static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
MpegEncContext * const s= (MpegEncContext *)c;
- DCTELEM temp[64];
+ uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+ DCTELEM * const temp= (DCTELEM*)aligned_temp;
int sum=0, i;
s->dsp.diff_pixels(temp, src1, src2, stride);
- s->fdct(temp);
+ s->dsp.fdct(temp);
for(i=0; i<64; i++)
sum+= ABS(temp[i]);
return sum;
}
-void simple_idct(INT16 *block); //FIXME
+void simple_idct(DCTELEM *block); //FIXME
static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
MpegEncContext * const s= (MpegEncContext *)c;
- DCTELEM temp[64], bak[64];
+ 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;
s->mb_intra=0;
memcpy(bak, temp, 64*sizeof(DCTELEM));
- s->dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
+ s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
s->dct_unquantize(s, temp, 0, s->qscale);
simple_idct(temp); //FIXME
return sum;
}
+static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+ 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;
+
+ 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){
+ s->dct_unquantize(s, temp, 0, s->qscale);
+ }
+
+ s->dsp.idct_add(bak, stride, temp);
+
+ distoration= s->dsp.sse[1](NULL, bak, src1, stride);
+
+ 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){
+ 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;
+
+ 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;
+}
+
+
WARPER88_1616(hadamard8_diff_c, hadamard8_diff16_c)
WARPER88_1616(dct_sad8x8_c, dct_sad16x16_c)
WARPER88_1616(quant_psnr8x8_c, quant_psnr16x16_c)
+WARPER88_1616(rd8x8_c, rd16x16_c)
+WARPER88_1616(bit8x8_c, bit16x16_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)
+{
+ 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(DSPContext* c, unsigned mask)
+/* init static data */
+void dsputil_static_init(void)
{
- static int init_done = 0;
int i;
- if (!init_done) {
- for(i=0;i<256;i++) cropTbl[i + MAX_NEG_CROP] = i;
- for(i=0;i<MAX_NEG_CROP;i++) {
- cropTbl[i] = 0;
- cropTbl[i + MAX_NEG_CROP + 256] = 255;
- }
+ for(i=0;i<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;
+}
- 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;
+void dsputil_init(DSPContext* c, AVCodecContext *avctx)
+{
+ int i;
- init_done = 1;
+#ifdef CONFIG_ENCODERS
+ if(avctx->dct_algo==FF_DCT_FASTINT)
+ c->fdct = fdct_ifast;
+ else
+ c->fdct = ff_jpeg_fdct_islow; //slow/accurate/default
+#endif //CONFIG_ENCODERS
+
+ 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->get_pixels = get_pixels_c;
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_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; \
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->put_mspel_pixels_tab[0]= put_mspel8_mc00_c;
c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
c->put_mspel_pixels_tab[5]= put_mspel8_mc12_c;
c->put_mspel_pixels_tab[6]= put_mspel8_mc22_c;
c->put_mspel_pixels_tab[7]= put_mspel8_mc32_c;
-
+
c->hadamard8_diff[0]= hadamard8_diff16_c;
c->hadamard8_diff[1]= hadamard8_diff_c;
c->hadamard8_abs = hadamard8_abs_c;
c->quant_psnr[0]= quant_psnr16x16_c;
c->quant_psnr[1]= quant_psnr8x8_c;
-
+
+ c->rd[0]= rd16x16_c;
+ c->rd[1]= rd8x8_c;
+
+ c->bit[0]= bit16x16_c;
+ c->bit[1]= bit8x8_c;
+
c->add_bytes= add_bytes_c;
c->diff_bytes= diff_bytes_c;
+ c->bswap_buf= bswap_buf;
#ifdef HAVE_MMX
- dsputil_init_mmx(c, mask);
- if (ff_bit_exact)
- {
- /* FIXME - AVCodec context should have flag for bitexact match */
- /* fprintf(stderr, "\n\n\nff_bit_exact %d\n\n\n\n", ff_bit_exact); */
- dsputil_set_bit_exact_mmx(c, mask);
- }
+ dsputil_init_mmx(c, avctx);
#endif
#ifdef ARCH_ARMV4L
- dsputil_init_armv4l(c, mask);
+ dsputil_init_armv4l(c, avctx);
#endif
#ifdef HAVE_MLIB
- dsputil_init_mlib(c, mask);
+ dsputil_init_mlib(c, avctx);
#endif
#ifdef ARCH_ALPHA
- dsputil_init_alpha(c, mask);
+ dsputil_init_alpha(c, avctx);
#endif
#ifdef ARCH_POWERPC
- dsputil_init_ppc(c, mask);
+ dsputil_init_ppc(c, avctx);
#endif
#ifdef HAVE_MMI
- dsputil_init_mmi(c, mask);
+ dsputil_init_mmi(c, avctx);
#endif
-}
-
-/* remove any non bit exact operation (testing purpose) */
-void avcodec_set_bit_exact(void)
-{
- ff_bit_exact=1;
-#ifdef HAVE_MMX
-// FIXME - better set_bit_exact
-// dsputil_set_bit_exact_mmx();
+#ifdef ARCH_SH4
+ dsputil_init_sh4(c,avctx);
#endif
+
+ 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:
+ fprintf(stderr, "Internal error, IDCT permutation not set\n");
+ }
}
+