x264 - predict.c

/****************************************************************************

* 16x16 prediction for intra luma block

****************************************************************************/

#define PREDICT_16x16_DC(v)\

for( int i = 0; i < 16; i++ )\

{\

MPIXEL_X4( src+ 0 ) = v;\

MPIXEL_X4( src+ 4 ) = v;\

MPIXEL_X4( src+ 8 ) = v;\

MPIXEL_X4( src+12 ) = v;\

src += FDEC_STRIDE;\

}

void x264_predict_16x16_dc_c( pixel *src )

{

int dc = 0;

/*

dc = H0,-1 + ... + H15,-1 + V-1,0 + ... + V-1,15

dc = (dc + 16） / 32

用dc组合成一个32bits字（即dcsplat的第一个字节是dc, 第二个字节也是dc, 第三个字节也是dc, 第四个字节也是dc）

PREDICT_16x16_DC给16行赋值，每行按4个32bits字赋值，都赋值为dcsplat;

*/

for( int i = 0; i < 16; i++ )

{

dc += src[-1 + i * FDEC_STRIDE];

dc += src[i - FDEC_STRIDE];

}

pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 16 ) >> 5 );

PREDICT_16x16_DC( dcsplat );

}

static void x264_predict_16x16_dc_left_c( pixel *src )

{

int dc = 0;

/*

dc = V-1,0 + ... + V-1,15

dc = (dc + 8) / 16

用dc组合成一个32bits字（即dcsplat的第一个字节是dc, 第二个字节也是dc, 第三个字节也是dc, 第四个字节也是dc）

PREDICT_16x16_DC给16行赋值，每行按4个32bits字赋值，都赋值为dcsplat;

*/

for( int i = 0; i < 16; i++ )

dc += src[-1 + i * FDEC_STRIDE];

pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );

PREDICT_16x16_DC( dcsplat );

}

static void x264_predict_16x16_dc_top_c( pixel *src )

{

int dc = 0;

/*

dc = H0,-1 + ... + H15,-1

dc = (dc + 8) / 16

用dc组合成一个32bits字（即dcsplat的第一个字节是dc, 第二个字节也是dc, 第三个字节也是dc, 第四个字节也是dc）

PREDICT_16x16_DC给16行赋值，每行按4个32bits字赋值，都赋值为dcsplat;

*/

for( int i = 0; i < 16; i++ )

dc += src[i - FDEC_STRIDE];

pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 );

PREDICT_16x16_DC( dcsplat );

}

static void x264_predict_16x16_dc_128_c( pixel *src )

{

/*

将16行16列的每个字节赋值为128

*/

PREDICT_16x16_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );

}

void x264_predict_16x16_h_c( pixel *src )

{

/*

将src[-1]字节，延展为32bits字v, v的每个字节都是src[-1]

然后将每一行的4个32bits字都赋值为v

*/

for( int i = 0; i < 16; i++ )

{

const pixel4 v = PIXEL_SPLAT_X4( src[-1] );

MPIXEL_X4( src+ 0 ) = v;

MPIXEL_X4( src+ 4 ) = v;

MPIXEL_X4( src+ 8 ) = v;

MPIXEL_X4( src+12 ) = v;

src += FDEC_STRIDE;

}

}

void x264_predict_16x16_v_c( pixel *src )

{

/*

将 -1 行的4个32bits的字分别赋值给v0, v1, v2, v3

将16行中的每一行的4个32bits字， 分别赋值为v0, v1, v2, v3

*/

pixel4 v0 = MPIXEL_X4( &src[ 0-FDEC_STRIDE] );

pixel4 v1 = MPIXEL_X4( &src[ 4-FDEC_STRIDE] );

pixel4 v2 = MPIXEL_X4( &src[ 8-FDEC_STRIDE] );

pixel4 v3 = MPIXEL_X4( &src[12-FDEC_STRIDE] );

for( int i = 0; i < 16; i++ )

{

MPIXEL_X4( src+ 0 ) = v0;

MPIXEL_X4( src+ 4 ) = v1;

MPIXEL_X4( src+ 8 ) = v2;

MPIXEL_X4( src+12 ) = v3;

src += FDEC_STRIDE;

}

}

void x264_predict_16x16_p_c( pixel *src )

{

int H = 0, V = 0;

/* calculate H and V */

/*

H = H8,-1 - H6,-1 + 2(H9,-1 - H5,-1) + 3(H10,-1 - H4,-1) + 4(H11,-1 - H3,-1)

+ 5(H12,-1 - H2,-1) + 6(H13,-1 - H1,-1) + 7(H14,-1 - H0,-1) + 8(H15,-1 - H-1,-1)

V = V-1,8 - V-1,6 + 2(V-1,9 - V-1,5) +3(V-1,10 - V-1,4) + 4(V-1,11 - V-1,3)

+ 5(V-1,12 - V-1,2) + 6(V-1,13 - V-1,1) + 7(V-1,14 - V-1,0) + 8(V-1,15 - V-1,-1)

H + V = H8,-1 + V-1,8 - (H6,-1 + V-1,6)

+ 2(H9,-1 + V-1,9) - 2(H5,-1 + V-1,5)

+ 3(H10,-1 + V-1,10) - 3(H4,-1 + V-1,4)

+ 4(H11,-1 + V-1,11) - 4(H3,-1 + V-1,3)

+ 5(H12,-1 + V-1,12) - 5(H2,-1 + V-1,2)

+ 6(H13,-1 + V-1,13) - 6(H1,-1 + V-1,1)

+ 7(H14,-1 + V-1,14) - 7(H0,-1 + V-1,0)

+ 8(H15,-1 + V-1,15) - 8(H-1,-1 + V-1,-1)

*/

for( int i = 0; i <= 7; i++ )

{

H += ( i + 1 ) * ( src[ 8 + i - FDEC_STRIDE ] - src[6 -i -FDEC_STRIDE] );

V += ( i + 1 ) * ( src[-1 + (8+i)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );

}

/*

a = 16(H15,-1 + V-1,15)

b = (5H + 32) / 64

c = (5V + 32) / 64

*/

int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] );

int b = ( 5 * H + 32 ) >> 6;

int c = ( 5 * V + 32 ) >> 6;

int i00 = a - b * 7 - c * 7 + 16;

for( int y = 0; y < 16; y++ )

{

int pix = i00;

for( int x = 0; x < 16; x++ )

{

src[x] = x264_clip_pixel( pix>>5 );

pix += b;

}

src += FDEC_STRIDE;

i00 += c;

}

}

/****************************************************************************

* 8x8 prediction for intra chroma block (4:2:0)

****************************************************************************/

static void x264_predict_8x8c_dc_128_c( pixel *src )

{

/*

将每一行的每个字节都赋值为128

*/

for( int y = 0; y < 8; y++ )

{

MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );

MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );

src += FDEC_STRIDE;

}

}

static void x264_predict_8x8c_dc_left_c( pixel *src )

{

int dc0 = 0, dc1 = 0;

/*

dc0 = V-1,0 + ... + V-1,3

dc1 = V-1,4 + ... + V-1,7

dc0 = (dc0 + 2) / 4;

dc1 = (dc1 + 2) / 4;

dc0splat = dc0延展为32bits字， 每个字节=dc0

dc0splat = dc1延展为32bits字， 每个字节=dc1

将 0 - 3 行的每一行的2个32bits字都赋值为dc0splat

将 4 - 7 行的每一行的2个32bits字都赋值为dc1splat

*/

for( int y = 0; y < 4; y++ )

{

dc0 += src[y * FDEC_STRIDE - 1];

dc1 += src[(y+4) * FDEC_STRIDE - 1];

}

pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );

pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc0splat;

MPIXEL_X4( src+4 ) = dc0splat;

src += FDEC_STRIDE;

}

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc1splat;

MPIXEL_X4( src+4 ) = dc1splat;

src += FDEC_STRIDE;

}

}

static void x264_predict_8x8c_dc_top_c( pixel *src )

{

int dc0 = 0, dc1 = 0;

/*

dc0 = H0,-1 + ... + H3,-1

dc1 = H4,-1 + ... + H7,-1

dc0 = (dc0 + 2) / 4;

dc1 = (dc1 + 2) / 4;

dc0splat = dc0延展为32bits字， 每个字节=dc0

dc0splat = dc1延展为32bits字， 每个字节=dc1

将 0 - 7 行的每一行的2个32bits字的第一个字赋值为dc0splat， 第二个字赋值为dc1splat

*/

for( int x = 0; x < 4; x++ )

{

dc0 += src[x - FDEC_STRIDE];

dc1 += src[x + 4 - FDEC_STRIDE];

}

pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );

pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );

for( int y = 0; y < 8; y++ )

{

MPIXEL_X4( src+0 ) = dc0splat;

MPIXEL_X4( src+4 ) = dc1splat;

src += FDEC_STRIDE;

}

}

void x264_predict_8x8c_dc_c( pixel *src )

{

int s0 = 0, s1 = 0, s2 = 0, s3 = 0;

/*

s0 = H0,-1 + ... + H3,-1

s1 = H4,-1 + ... + H7,-1

s2 = V-1,0 + ... + V-1,3

s3 = V-1,4 + ... + V-1,7

*/

for( int i = 0; i < 4; i++ )

{

s0 += src[i - FDEC_STRIDE];

s1 += src[i + 4 - FDEC_STRIDE];

s2 += src[-1 + i * FDEC_STRIDE];

s3 += src[-1 + (i+4)*FDEC_STRIDE];

}

/*

dc0 = (s0 + s2 + 4) / 8， dc0 延展

dc1 = (s1 + 2) / 4， dc1 延展

dc2 = (s3 + 2) / 4， dc2 延展

dc3 = (s1 + s3 + 4) / 8， dc3 延展

*/

pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );

pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );

pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );

pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );

/*

将 0 - 3行的2个32bits字分别赋值为dc0, dc1

*/

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc0;

MPIXEL_X4( src+4 ) = dc1;

src += FDEC_STRIDE;

}

/*

将 4 - 7行的2个32bits字分别赋值为dc2, dc3

*/

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc2;

MPIXEL_X4( src+4 ) = dc3;

src += FDEC_STRIDE;

}

}

void x264_predict_8x8c_h_c( pixel *src )

{

/*

将每一行的src[-1] 赋值给这行的每一个字节

*/

for( int i = 0; i < 8; i++ )

{

pixel4 v = PIXEL_SPLAT_X4( src[-1] );

MPIXEL_X4( src+0 ) = v;

MPIXEL_X4( src+4 ) = v;

src += FDEC_STRIDE;

}

}

void x264_predict_8x8c_v_c( pixel *src )

{

/*

将 -1 行 的 2 个 32bits的字 分别 赋值给 v0, v1

将8行中的每一行的2个32bits字分别赋值为v0, v1

*/

pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );

pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );

for( int i = 0; i < 8; i++ )

{

MPIXEL_X4( src+0 ) = v0;

MPIXEL_X4( src+4 ) = v1;

src += FDEC_STRIDE;

}

}

void x264_predict_8x8c_p_c( pixel *src )

{

int H = 0, V = 0;

/*

H = H4,-1 - H2,-1 + 2(H5,-1 - H1,-1) + 3(H6,-1 - H0,-1) + 4(H7,-1 - H-1,-1)

V = V-1,4 - V-1,2 + 2(V-1,5 - V-1,1) + 3(V-1,6 - V-1,0) + 4(V-1,7 - V-1,-1)

*/

for( int i = 0; i < 4; i++ )

{

H += ( i + 1 ) * ( src[4+i - FDEC_STRIDE] - src[2 - i -FDEC_STRIDE] );

V += ( i + 1 ) * ( src[-1 +(i+4)*FDEC_STRIDE] - src[-1+(2-i)*FDEC_STRIDE] );

}

/*

a = 16(V-1,7 + H7,-1)

b = (17H + 16) / 32

c = (17V + 16) / 32

i00 = a - 3b - 3c + 16

*/

int a = 16 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );

int b = ( 17 * H + 16 ) >> 5;

int c = ( 17 * V + 16 ) >> 5;

int i00 = a -3*b -3*c + 16;

for( int y = 0; y < 8; y++ )

{

int pix = i00;

for( int x = 0; x < 8; x++ )

{

src[x] = x264_clip_pixel( pix>>5 );

pix += b;

}

src += FDEC_STRIDE;

i00 += c;

}

}

/****************************************************************************

* 8x16 prediction for intra chroma block (4:2:2)

****************************************************************************/

static void x264_predict_8x16c_dc_128_c( pixel *src )

{

/*

将16行的每一行的每一个字节都赋值为128

*/

for( int y = 0; y < 16; y++ )

{

MPIXEL_X4( src+0 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );

MPIXEL_X4( src+4 ) = PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) );

src += FDEC_STRIDE;

}

}

static void x264_predict_8x16c_dc_left_c( pixel *src )

{

/*

将16行以4行为一个单位， 对这个四行求dc

dc = V-1,0 + ... + V-1,3

dc = (dc + 2) / 4

dc延展为dcsplat

将dcsplat赋值给这四行中的每一行

*/

for( int i = 0; i < 4; i++ )

{

int dc = 0;

for( int y = 0; y < 4; y++ )

dc += src[y*FDEC_STRIDE - 1];

pixel4 dcsplat = PIXEL_SPLAT_X4( (dc + 2) >> 2 );

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dcsplat;

MPIXEL_X4( src+4 ) = dcsplat;

src += FDEC_STRIDE;

}

}

}

static void x264_predict_8x16c_dc_top_c( pixel *src )

{

int dc0 = 0, dc1 = 0;

/*

dc0= H0,-1 + ... + H3,-1

dc1= H4,-1 + ... + H7,-1

dc0 = (dc0 + 2) / 4

dc1 = (dc1 + 2) / 4

dc0延展为dc0splat, dc1 延展为dc1splat

将dcsplat0, dcsplat1赋值给16行中的每一行

*/

for(int x = 0; x < 4; x++ )

{

dc0 += src[x - FDEC_STRIDE];

dc1 += src[x + 4 - FDEC_STRIDE];

}

pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 );

pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 );

for( int y = 0; y < 16; y++ )

{

MPIXEL_X4( src+0 ) = dc0splat;

MPIXEL_X4( src+4 ) = dc1splat;

src += FDEC_STRIDE;

}

}

void x264_predict_8x16c_dc_c( pixel *src )

{

int s0 = 0, s1 = 0, s2 = 0, s3 = 0, s4 = 0, s5 = 0;

/*

s0 s1

s2

s3

s4

s5

*/

/*

s0 = H0,-1 + ... + H3,-1

s1 = H4,-1 + ... + H7,-1

s2 = V-1,0 + ... + V-1,3

s3 = V-1,4 + ... + V-1,7

s4 = V-1,8 + ... + V-1,11

s5 = V-1,12 + ... + V-1,15,

*/

for( int i = 0; i < 4; i++ )

{

s0 += src[i+0 - FDEC_STRIDE];

s1 += src[i+4 - FDEC_STRIDE];

s2 += src[-1 + (i+0) * FDEC_STRIDE];

s3 += src[-1 + (i+4) * FDEC_STRIDE];

s4 += src[-1 + (i+8) * FDEC_STRIDE];

s5 += src[-1 + (i+12) * FDEC_STRIDE];

}

/*

dc0 dc1

dc2 dc3

dc4 dc5

dc6 dc7

*/

/*

dc0 = (s0 + s2 + 4) / 8

dc1 = (s1 + 2) / 4

dc2 = (s3 + 2) / 4

dc3 = (s1 + s3 + 4) / 8

dc4 = (s4 + 2) / 4

dc5 = (s1 + s4 + 4) / 8

dc6 = (s5 + 2) / 4

dc7 = (s1 + s5 + 4) / 8

将dc0延展， dc1延展， 然后赋值给0-3行

dc2, dc3分别延展， 然后赋值给4-7行

dc4, dc5分别延展， 然后赋值给8-11行

dc6, dc7分别延展， 然后赋值给12-15行

*/

pixel4 dc0 = PIXEL_SPLAT_X4( ( s0 + s2 + 4 ) >> 3 );

pixel4 dc1 = PIXEL_SPLAT_X4( ( s1 + 2 ) >> 2 );

pixel4 dc2 = PIXEL_SPLAT_X4( ( s3 + 2 ) >> 2 );

pixel4 dc3 = PIXEL_SPLAT_X4( ( s1 + s3 + 4 ) >> 3 );

pixel4 dc4 = PIXEL_SPLAT_X4( ( s4 + 2 ) >> 2 );

pixel4 dc5 = PIXEL_SPLAT_X4( ( s1 + s4 + 4 ) >> 3 );

pixel4 dc6 = PIXEL_SPLAT_X4( ( s5 + 2 ) >> 2 );

pixel4 dc7 = PIXEL_SPLAT_X4( ( s1 + s5 + 4 ) >> 3 );

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc0;

MPIXEL_X4( src+4 ) = dc1;

src += FDEC_STRIDE;

}

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc2;

MPIXEL_X4( src+4 ) = dc3;

src += FDEC_STRIDE;

}

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc4;

MPIXEL_X4( src+4 ) = dc5;

src += FDEC_STRIDE;

}

for( int y = 0; y < 4; y++ )

{

MPIXEL_X4( src+0 ) = dc6;

MPIXEL_X4( src+4 ) = dc7;

src += FDEC_STRIDE;

}

}

void x264_predict_8x16c_h_c( pixel *src )

{

/*

将每一行的src[-1] 赋值给这行的每一个字节

*/

for( int i = 0; i < 16; i++ )

{

pixel4 v = PIXEL_SPLAT_X4( src[-1] );

MPIXEL_X4( src+0 ) = v;

MPIXEL_X4( src+4 ) = v;

src += FDEC_STRIDE;

}

}

void x264_predict_8x16c_v_c( pixel *src )

{

pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE );

pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE );

/*

将 -1 行的 2 个 32 bits 的 字 分别赋值给v0， v1

然后将v0, v1赋值给16行中的每一行

*/

for( int i = 0; i < 16; i++ )

{

MPIXEL_X4( src+0 ) = v0;

MPIXEL_X4( src+4 ) = v1;

src += FDEC_STRIDE;

}

}

void x264_predict_8x16c_p_c( pixel *src )

{

int H = 0;

int V = 0;

/*

H = H4,-1 - H2,-1 + 2(H5,-1 - H1,-1) + 3(H6,-1 - H0,-1) + 4(H7,-1 - H-1,-1)

V = V-1,8 - V-1,6 + 2(V-1,9 - V-1,5) + 3(V-1,10 - V-1,4) + 4(V-1,11 - V-1,3) + 5(V-1,12 - V-1,2) + 6(V-1,13 - V-1,1) + 7(V-1,14 - V-1,0) + 8(V-1,15 - V-1,-1)

*/

for( int i = 0; i < 4; i++ )

H += ( i + 1 ) * ( src[4 + i - FDEC_STRIDE] - src[2 - i - FDEC_STRIDE] );

for( int i = 0; i < 8; i++ )

V += ( i + 1 ) * ( src[-1 + (i+8)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );

/*

a = 16(H7,-1 + V-1,15)

b = (17H + 16) / 32

c = (5V + 32) / 64

i00 = a - 3b -7c + 16

*/

int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );

int b = ( 17 * H + 16 ) >> 5;

int c = ( 5 * V + 32 ) >> 6;

int i00 = a -3*b -7*c + 16;

for( int y = 0; y < 16; y++ )

{

int pix = i00;

for( int x = 0; x < 8; x++ )

{

src[x] = x264_clip_pixel( pix>>5 );

pix += b;

}

src += FDEC_STRIDE;

i00 += c;

}

}

/****************************************************************************

* 4x4 prediction for intra luma block

****************************************************************************/

#define SRC(x,y) src[(x)+(y)*FDEC_STRIDE] // src[y][x]

#define SRC_X4(x,y) MPIXEL_X4( &SRC(x,y) ) // (uint32_t *)&src[y][x]

// SRC_X4(0, 0) denote 0 row of 4x4

// SRC_X4(0, 1) denote 1 row of 4x4

// SRC_X4(0, 2) denote 2 row of 4x4

// SRC_X4(0, 3) denote 3 row of 4x4

#define PREDICT_4x4_DC(v)\

SRC_X4(0,0) = SRC_X4(0,1) = SRC_X4(0,2) = SRC_X4(0,3) = v;

// mode 2 (DC), but H and V both no available

static void x264_predict_4x4_dc_128_c( pixel *src )

{

PREDICT_4x4_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) ); // all = 128

}

// mode 2(DC), but H no available

static void x264_predict_4x4_dc_left_c( pixel *src )

{

pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2 );

PREDICT_4x4_DC( dc );

}

// mode 2(DC), but V no available

static void x264_predict_4x4_dc_top_c( pixel *src )

{

pixel4 dc = PIXEL_SPLAT_X4( (SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2 );

PREDICT_4x4_DC( dc );

}

// mode 2(DC), H and V both available

void x264_predict_4x4_dc_c( pixel *src )

{

pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) +

SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 4) >> 3 );

PREDICT_4x4_DC( dc );

}

// mode 1 (Horizon), take left byte and splat_x4 to 4 byte, then copy current row

void x264_predict_4x4_h_c( pixel *src )

{

SRC_X4(0,0) = PIXEL_SPLAT_X4( SRC(-1,0) );

SRC_X4(0,1) = PIXEL_SPLAT_X4( SRC(-1,1) );

SRC_X4(0,2) = PIXEL_SPLAT_X4( SRC(-1,2) );

SRC_X4(0,3) = PIXEL_SPLAT_X4( SRC(-1,3) );

}

// mode 0 (Vertical), take top row and copy each of 4x4 block

void x264_predict_4x4_v_c( pixel *src )

{

PREDICT_4x4_DC(SRC_X4(0,-1));

}

// take left pixel to l0, l1 etc

#define PREDICT_4x4_LOAD_LEFT\

int l0 = SRC(-1,0);\

int l1 = SRC(-1,1);\

int l2 = SRC(-1,2);\

UNUSED int l3 = SRC(-1,3);

// take top pixel to t0, t1 etc

#define PREDICT_4x4_LOAD_TOP\

int t0 = SRC(0,-1);\

int t1 = SRC(1,-1);\

int t2 = SRC(2,-1);\

UNUSED int t3 = SRC(3,-1);

// take top right to t4, t5 etc

#define PREDICT_4x4_LOAD_TOP_RIGHT\

int t4 = SRC(4,-1);\

int t5 = SRC(5,-1);\

int t6 = SRC(6,-1);\

UNUSED int t7 = SRC(7,-1);

// mean of a, b, then round

#define F1(a,b) (((a)+(b)+1)>>1)

// weighted mean of a, b, c, then round

#define F2(a,b,c) (((a)+2*(b)+(c)+2)>>2)

// mode 3 (Dialog Down Left)

static void x264_predict_4x4_ddl_c( pixel *src )

{

PREDICT_4x4_LOAD_TOP

PREDICT_4x4_LOAD_TOP_RIGHT

SRC(0,0)= F2(t0,t1,t2); // src[0][0] = Round(t0, t1, t2's weighted mean)

SRC(1,0)=SRC(0,1)= F2(t1,t2,t3); // src[0][1] = src[1][0] = Round(t1, t2, t3's weighted mean)

SRC(2,0)=SRC(1,1)=SRC(0,2)= F2(t2,t3,t4);

SRC(3,0)=SRC(2,1)=SRC(1,2)=SRC(0,3)= F2(t3,t4,t5);

SRC(3,1)=SRC(2,2)=SRC(1,3)= F2(t4,t5,t6);

SRC(3,2)=SRC(2,3)= F2(t5,t6,t7);

SRC(3,3)= F2(t6,t7,t7);

}

// mode 4 (Dialog Down Right)

static void x264_predict_4x4_ddr_c( pixel *src )

{

int lt = SRC(-1,-1);

PREDICT_4x4_LOAD_LEFT

PREDICT_4x4_LOAD_TOP

SRC(3,0)= F2(t3,t2,t1); // assign src[0][3]

SRC(2,0)=SRC(3,1)= F2(t2,t1,t0); // assign src[0][2], src[1][3

SRC(1,0)=SRC(2,1)=SRC(3,2)= F2(t1,t0,lt);

SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)= F2(t0,lt,l0);

SRC(0,1)=SRC(1,2)=SRC(2,3)= F2(lt,l0,l1);

SRC(0,2)=SRC(1,3)= F2(l0,l1,l2);

SRC(0,3)= F2(l1,l2,l3);

}

// mode 5 (Vertical Right)

static void x264_predict_4x4_vr_c( pixel *src )

{

int lt = SRC(-1,-1);

PREDICT_4x4_LOAD_LEFT

PREDICT_4x4_LOAD_TOP

SRC(0,3)= F2(l2,l1,l0); // assign src[3][0], weighted mean as l1 center rather than dialog l2 as center

SRC(0,2)= F2(l1,l0,lt); // assign src[2][0], weighted mean as l0 center.

SRC(0,1)=SRC(1,3)= F2(l0,lt,t0); // assign src[1][0], src[3][1], weigthed mean as lt as center

SRC(0,0)=SRC(1,2)= F1(lt,t0); // assign src[0][0], src[2][1], mean of lt, t0

SRC(1,1)=SRC(2,3)= F2(lt,t0,t1); // assign src[1][1], src[3][2], t0 as center

SRC(1,0)=SRC(2,2)= F1(t0,t1); // src[0][1], src[2][2]; mean of t0, t1

SRC(2,1)=SRC(3,3)= F2(t0,t1,t2); // src[1][2], src[3][3]; t1 as center

SRC(2,0)=SRC(3,2)= F1(t1,t2); // src[0][2], src[2][3]; mean of t1, t2

SRC(3,1)= F2(t1,t2,t3); // src[1][3]; t2 as center

SRC(3,0)= F1(t2,t3); // src[0][3]; mean of t2, t3

}

// mode 6 (Horizon Down)

static void x264_predict_4x4_hd_c( pixel *src )

{

int lt= SRC(-1,-1);

PREDICT_4x4_LOAD_LEFT

PREDICT_4x4_LOAD_TOP

SRC(0,3)= F1(l2,l3); // src[3][0]; mean of l2, l3

SRC(1,3)= F2(l1,l2,l3); // src[3][1]; as l2 center

SRC(0,2)=SRC(2,3)= F1(l1,l2); // src[2][0], src[3][2]; mean of l1, l2

SRC(1,2)=SRC(3,3)= F2(l0,l1,l2); // src[2][1], src[3][3]; as l1 as center

SRC(0,1)=SRC(2,2)= F1(l0,l1); // src[1][0], src[2][2]; mean of l0, l1

SRC(1,1)=SRC(3,2)= F2(lt,l0,l1); // src[1][1], src[2][3]; as l0 as center

SRC(0,0)=SRC(2,1)= F1(lt,l0); // src[0][0], src[1][2]; mean of lt, l0

SRC(1,0)=SRC(3,1)= F2(t0,lt,l0); // src[0][1], src[1][3]; as lt center

SRC(2,0)= F2(t1,t0,lt); // src[0][2]; as t0 center

SRC(3,0)= F2(t2,t1,t0); // src[0][3]; as t1 center

}

// mode 7 (Vertical Left)

static void x264_predict_4x4_vl_c( pixel *src )

{

PREDICT_4x4_LOAD_TOP

PREDICT_4x4_LOAD_TOP_RIGHT

SRC(0,0)= F1(t0,t1); // src[0][0]; mean t0, t1

SRC(0,1)= F2(t0,t1,t2); // src[1][0]; as t1 center

SRC(1,0)=SRC(0,2)= F1(t1,t2); // src[0][1], src[2][0]; mean of t1, t2

SRC(1,1)=SRC(0,3)= F2(t1,t2,t3); // src[1][1], src[3][0]; as t2 center

SRC(2,0)=SRC(1,2)= F1(t2,t3); // src[0][2], src[2][1]; mean of t2, t3

SRC(2,1)=SRC(1,3)= F2(t2,t3,t4); // src[1][2], src[3][1]; as t3 center

SRC(3,0)=SRC(2,2)= F1(t3,t4); // src[0][3], src[2][2]; mean of t3, t4

SRC(3,1)=SRC(2,3)= F2(t3,t4,t5); // src[1][3], src[3][2]; t4 as center

SRC(3,2)= F1(t4,t5); // src[2][3]; mean of t4, t5

SRC(3,3)= F2(t4,t5,t6); // src[3][3]; as t5 center

}

// mode 8 (horizon up)

static void x264_predict_4x4_hu_c( pixel *src )

{

PREDICT_4x4_LOAD_LEFT

SRC(0,0)= F1(l0,l1); // src[0][0]; mean of l0, l1

SRC(1,0)= F2(l0,l1,l2); // src[0][1]; as l1 center

SRC(2,0)=SRC(0,1)= F1(l1,l2); // src[0][2], src[1][0]; mean of l1, l2

SRC(3,0)=SRC(1,1)= F2(l1,l2,l3); // src[0][3], src[1][1]; l2 as center

SRC(2,1)=SRC(0,2)= F1(l2,l3); // src[1][2], src[2][0]; mean of l2, l3

SRC(3,1)=SRC(1,2)= F2(l2,l3,l3); // src[1][3], src[2][1]; as l3 center

SRC(3,2)=SRC(1,3)=SRC(0,3)=

SRC(2,2)=SRC(2,3)=SRC(3,3)= l3; // others = l3

}

/****************************************************************************

* 8x8 prediction for intra luma block

****************************************************************************/

#define PL(y) \

edge[14-y] = F2(SRC(-1,y-1), SRC(-1,y), SRC(-1,y+1));

#define PT(x) \

edge[16+x] = F2(SRC(x-1,-1), SRC(x,-1), SRC(x+1,-1));

static void x264_predict_8x8_filter_c( pixel *src, pixel edge[36], int i_neighbor, int i_filters )

{

/* edge[7..14] = l7..l0

* edge[15] = lt

* edge[16..31] = t0 .. t15

* edge[32] = t15 */

int have_lt = i_neighbor & MB_TOPLEFT;

if( i_filters & MB_LEFT )

{

edge[15] = (SRC(0,-1) + 2*SRC(-1,-1) + SRC(-1,0) + 2) >> 2;

edge[14] = ((have_lt ? SRC(-1,-1) : SRC(-1,0))

+ 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2;

PL(1) PL(2) PL(3) PL(4) PL(5) PL(6)

edge[6] =

edge[7] = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2;

}

if( i_filters & MB_TOP )

{

int have_tr = i_neighbor & MB_TOPRIGHT;

edge[16] = ((have_lt ? SRC(-1,-1) : SRC(0,-1))

+ 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2;

PT(1) PT(2) PT(3) PT(4) PT(5) PT(6)

edge[23] = (SRC(6,-1) + 2*SRC(7,-1)

+ (have_tr ? SRC(8,-1) : SRC(7,-1)) + 2) >> 2;

if( i_filters & MB_TOPRIGHT )

{

if( have_tr )

{

PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14)

edge[31] =

edge[32] = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2;

}

else

{

MPIXEL_X4( edge+24 ) = PIXEL_SPLAT_X4( SRC(7,-1) );

MPIXEL_X4( edge+28 ) = PIXEL_SPLAT_X4( SRC(7,-1) );

edge[32] = SRC(7,-1);

}

}

}

}

#undef PL

#undef PT

#define PL(y) \

UNUSED int l##y = edge[14-y];

#define PT(x) \

UNUSED int t##x = edge[16+x];

#define PREDICT_8x8_LOAD_TOPLEFT \

int lt = edge[15];

#define PREDICT_8x8_LOAD_LEFT \

PL(0) PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) PL(7)

#define PREDICT_8x8_LOAD_TOP \

PT(0) PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) PT(7)

#define PREDICT_8x8_LOAD_TOPRIGHT \

PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14) PT(15)

#define PREDICT_8x8_DC(v) \

for( int y = 0; y < 8; y++ ) { \

MPIXEL_X4( src+0 ) = v; \

MPIXEL_X4( src+4 ) = v; \

src += FDEC_STRIDE; \

}

static void x264_predict_8x8_dc_128_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) );

}

static void x264_predict_8x8_dc_left_c( pixel *src, pixel edge[36] )

{

// load l0, ..., l7 from edge

PREDICT_8x8_LOAD_LEFT

// 取平均值v, 将v延展为四字节， 每个字节都为v, 将延展结果赋值给dc

pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3 );

// 将dc赋值给8行中的每一行的两个4字节字， 这样8x8块的每一字节都将是v

PREDICT_8x8_DC( dc );

}

static void x264_predict_8x8_dc_top_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

pixel4 dc = PIXEL_SPLAT_X4( (t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3 );

PREDICT_8x8_DC( dc );

}

void x264_predict_8x8_dc_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_LEFT

PREDICT_8x8_LOAD_TOP

pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4 );

PREDICT_8x8_DC( dc );

}

void x264_predict_8x8_h_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_LEFT

#define ROW(y) MPIXEL_X4( src+y*FDEC_STRIDE+0 ) =\

MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = PIXEL_SPLAT_X4( l##y );

ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);

#undef ROW

}

void x264_predict_8x8_v_c( pixel *src, pixel edge[36] )

{

pixel4 top[2] = { MPIXEL_X4( edge+16 ),

MPIXEL_X4( edge+20 ) };

for( int y = 0; y < 8; y++ )

{

MPIXEL_X4( src+y*FDEC_STRIDE+0 ) = top[0];

MPIXEL_X4( src+y*FDEC_STRIDE+4 ) = top[1];

}

}

static void x264_predict_8x8_ddl_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

PREDICT_8x8_LOAD_TOPRIGHT

SRC(0,0)= F2(t0,t1,t2);

SRC(0,1)=SRC(1,0)= F2(t1,t2,t3);

SRC(0,2)=SRC(1,1)=SRC(2,0)= F2(t2,t3,t4);

SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= F2(t3,t4,t5);

SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= F2(t4,t5,t6);

SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= F2(t5,t6,t7);

SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= F2(t6,t7,t8);

SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= F2(t7,t8,t9);

SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= F2(t8,t9,t10);

SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= F2(t9,t10,t11);

SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= F2(t10,t11,t12);

SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= F2(t11,t12,t13);

SRC(5,7)=SRC(6,6)=SRC(7,5)= F2(t12,t13,t14);

SRC(6,7)=SRC(7,6)= F2(t13,t14,t15);

SRC(7,7)= F2(t14,t15,t15);

}

static void x264_predict_8x8_ddr_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

PREDICT_8x8_LOAD_LEFT

PREDICT_8x8_LOAD_TOPLEFT

SRC(0,7)= F2(l7,l6,l5);

SRC(0,6)=SRC(1,7)= F2(l6,l5,l4);

SRC(0,5)=SRC(1,6)=SRC(2,7)= F2(l5,l4,l3);

SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= F2(l4,l3,l2);

SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= F2(l3,l2,l1);

SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= F2(l2,l1,l0);

SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= F2(l1,l0,lt);

SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= F2(l0,lt,t0);

SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= F2(lt,t0,t1);

SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= F2(t0,t1,t2);

SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= F2(t1,t2,t3);

SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= F2(t2,t3,t4);

SRC(5,0)=SRC(6,1)=SRC(7,2)= F2(t3,t4,t5);

SRC(6,0)=SRC(7,1)= F2(t4,t5,t6);

SRC(7,0)= F2(t5,t6,t7);

}

static void x264_predict_8x8_vr_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

PREDICT_8x8_LOAD_LEFT

PREDICT_8x8_LOAD_TOPLEFT

SRC(0,6)= F2(l5,l4,l3);

SRC(0,7)= F2(l6,l5,l4);

SRC(0,4)=SRC(1,6)= F2(l3,l2,l1);

SRC(0,5)=SRC(1,7)= F2(l4,l3,l2);

SRC(0,2)=SRC(1,4)=SRC(2,6)= F2(l1,l0,lt);

SRC(0,3)=SRC(1,5)=SRC(2,7)= F2(l2,l1,l0);

SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= F2(l0,lt,t0);

SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= F1(lt,t0);

SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= F2(lt,t0,t1);

SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= F1(t0,t1);

SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= F2(t0,t1,t2);

SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= F1(t1,t2);

SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= F2(t1,t2,t3);

SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= F1(t2,t3);

SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= F2(t2,t3,t4);

SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= F1(t3,t4);

SRC(5,1)=SRC(6,3)=SRC(7,5)= F2(t3,t4,t5);

SRC(5,0)=SRC(6,2)=SRC(7,4)= F1(t4,t5);

SRC(6,1)=SRC(7,3)= F2(t4,t5,t6);

SRC(6,0)=SRC(7,2)= F1(t5,t6);

SRC(7,1)= F2(t5,t6,t7);

SRC(7,0)= F1(t6,t7);

}

static void x264_predict_8x8_hd_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

PREDICT_8x8_LOAD_LEFT

PREDICT_8x8_LOAD_TOPLEFT

int p1 = pack_pixel_1to2(F1(l6,l7), F2(l5,l6,l7));

int p2 = pack_pixel_1to2(F1(l5,l6), F2(l4,l5,l6));

int p3 = pack_pixel_1to2(F1(l4,l5), F2(l3,l4,l5));

int p4 = pack_pixel_1to2(F1(l3,l4), F2(l2,l3,l4));

int p5 = pack_pixel_1to2(F1(l2,l3), F2(l1,l2,l3));

int p6 = pack_pixel_1to2(F1(l1,l2), F2(l0,l1,l2));

int p7 = pack_pixel_1to2(F1(l0,l1), F2(lt,l0,l1));

int p8 = pack_pixel_1to2(F1(lt,l0), F2(l0,lt,t0));

int p9 = pack_pixel_1to2(F2(t1,t0,lt), F2(t2,t1,t0));

int p10 = pack_pixel_1to2(F2(t3,t2,t1), F2(t4,t3,t2));

int p11 = pack_pixel_1to2(F2(t5,t4,t3), F2(t6,t5,t4));

SRC_X4(0,7)= pack_pixel_2to4(p1,p2);

SRC_X4(0,6)= pack_pixel_2to4(p2,p3);

SRC_X4(4,7)=SRC_X4(0,5)= pack_pixel_2to4(p3,p4);

SRC_X4(4,6)=SRC_X4(0,4)= pack_pixel_2to4(p4,p5);

SRC_X4(4,5)=SRC_X4(0,3)= pack_pixel_2to4(p5,p6);

SRC_X4(4,4)=SRC_X4(0,2)= pack_pixel_2to4(p6,p7);

SRC_X4(4,3)=SRC_X4(0,1)= pack_pixel_2to4(p7,p8);

SRC_X4(4,2)=SRC_X4(0,0)= pack_pixel_2to4(p8,p9);

SRC_X4(4,1)= pack_pixel_2to4(p9,p10);

SRC_X4(4,0)= pack_pixel_2to4(p10,p11);

}

static void x264_predict_8x8_vl_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_TOP

PREDICT_8x8_LOAD_TOPRIGHT

SRC(0,0)= F1(t0,t1);

SRC(0,1)= F2(t0,t1,t2);

SRC(0,2)=SRC(1,0)= F1(t1,t2);

SRC(0,3)=SRC(1,1)= F2(t1,t2,t3);

SRC(0,4)=SRC(1,2)=SRC(2,0)= F1(t2,t3);

SRC(0,5)=SRC(1,3)=SRC(2,1)= F2(t2,t3,t4);

SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= F1(t3,t4);

SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= F2(t3,t4,t5);

SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= F1(t4,t5);

SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= F2(t4,t5,t6);

SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= F1(t5,t6);

SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= F2(t5,t6,t7);

SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= F1(t6,t7);

SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= F2(t6,t7,t8);

SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= F1(t7,t8);

SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= F2(t7,t8,t9);

SRC(5,6)=SRC(6,4)=SRC(7,2)= F1(t8,t9);

SRC(5,7)=SRC(6,5)=SRC(7,3)= F2(t8,t9,t10);

SRC(6,6)=SRC(7,4)= F1(t9,t10);

SRC(6,7)=SRC(7,5)= F2(t9,t10,t11);

SRC(7,6)= F1(t10,t11);

SRC(7,7)= F2(t10,t11,t12);

}

static void x264_predict_8x8_hu_c( pixel *src, pixel edge[36] )

{

PREDICT_8x8_LOAD_LEFT

int p1 = pack_pixel_1to2(F1(l0,l1), F2(l0,l1,l2));

int p2 = pack_pixel_1to2(F1(l1,l2), F2(l1,l2,l3));

int p3 = pack_pixel_1to2(F1(l2,l3), F2(l2,l3,l4));

int p4 = pack_pixel_1to2(F1(l3,l4), F2(l3,l4,l5));

int p5 = pack_pixel_1to2(F1(l4,l5), F2(l4,l5,l6));

int p6 = pack_pixel_1to2(F1(l5,l6), F2(l5,l6,l7));

int p7 = pack_pixel_1to2(F1(l6,l7), F2(l6,l7,l7));

int p8 = pack_pixel_1to2(l7,l7);

SRC_X4(0,0)= pack_pixel_2to4(p1,p2);

SRC_X4(0,1)= pack_pixel_2to4(p2,p3);

SRC_X4(4,0)=SRC_X4(0,2)= pack_pixel_2to4(p3,p4);

SRC_X4(4,1)=SRC_X4(0,3)= pack_pixel_2to4(p4,p5);

SRC_X4(4,2)=SRC_X4(0,4)= pack_pixel_2to4(p5,p6);

SRC_X4(4,3)=SRC_X4(0,5)= pack_pixel_2to4(p6,p7);

SRC_X4(4,4)=SRC_X4(0,6)= pack_pixel_2to4(p7,p8);

SRC_X4(4,5)=SRC_X4(4,6)= SRC_X4(0,7) = SRC_X4(4,7) = pack_pixel_2to4(p8,p8);

}

/****************************************************************************

* Exported functions:

****************************************************************************/

void x264_predict_16x16_init( int cpu, x264_predict_t pf[7] )

{

pf[I_PRED_16x16_V ] = x264_predict_16x16_v_c;

pf[I_PRED_16x16_H ] = x264_predict_16x16_h_c;

pf[I_PRED_16x16_DC] = x264_predict_16x16_dc_c;

pf[I_PRED_16x16_P ] = x264_predict_16x16_p_c;

pf[I_PRED_16x16_DC_LEFT]= x264_predict_16x16_dc_left_c;

pf[I_PRED_16x16_DC_TOP ]= x264_predict_16x16_dc_top_c;

pf[I_PRED_16x16_DC_128 ]= x264_predict_16x16_dc_128_c;

#if HAVE_MMX

x264_predict_16x16_init_mmx( cpu, pf );

#endif

#if HAVE_ALTIVEC

if( cpu&X264_CPU_ALTIVEC )

x264_predict_16x16_init_altivec( pf );

#endif

#if HAVE_ARMV6

x264_predict_16x16_init_arm( cpu, pf );

#endif

}

void x264_predict_8x8c_init( int cpu, x264_predict_t pf[7] )

{

pf[I_PRED_CHROMA_V ] = x264_predict_8x8c_v_c;

pf[I_PRED_CHROMA_H ] = x264_predict_8x8c_h_c;

pf[I_PRED_CHROMA_DC] = x264_predict_8x8c_dc_c;

pf[I_PRED_CHROMA_P ] = x264_predict_8x8c_p_c;

pf[I_PRED_CHROMA_DC_LEFT]= x264_predict_8x8c_dc_left_c;

pf[I_PRED_CHROMA_DC_TOP ]= x264_predict_8x8c_dc_top_c;

pf[I_PRED_CHROMA_DC_128 ]= x264_predict_8x8c_dc_128_c;

#if HAVE_MMX

x264_predict_8x8c_init_mmx( cpu, pf );

#endif

#if HAVE_ALTIVEC

if( cpu&X264_CPU_ALTIVEC )

x264_predict_8x8c_init_altivec( pf );

#endif

#if HAVE_ARMV6

x264_predict_8x8c_init_arm( cpu, pf );

#endif

}

void x264_predict_8x16c_init( int cpu, x264_predict_t pf[7] )

{

pf[I_PRED_CHROMA_V ] = x264_predict_8x16c_v_c;

pf[I_PRED_CHROMA_H ] = x264_predict_8x16c_h_c;

pf[I_PRED_CHROMA_DC] = x264_predict_8x16c_dc_c;

pf[I_PRED_CHROMA_P ] = x264_predict_8x16c_p_c;

pf[I_PRED_CHROMA_DC_LEFT]= x264_predict_8x16c_dc_left_c;

pf[I_PRED_CHROMA_DC_TOP ]= x264_predict_8x16c_dc_top_c;

pf[I_PRED_CHROMA_DC_128 ]= x264_predict_8x16c_dc_128_c;

#if HAVE_MMX

x264_predict_8x16c_init_mmx( cpu, pf );

#endif

}

void x264_predict_8x8_init( int cpu, x264_predict8x8_t pf[12], x264_predict_8x8_filter_t *predict_filter )

{

pf[I_PRED_8x8_V] = x264_predict_8x8_v_c;

pf[I_PRED_8x8_H] = x264_predict_8x8_h_c;

pf[I_PRED_8x8_DC] = x264_predict_8x8_dc_c;

pf[I_PRED_8x8_DDL] = x264_predict_8x8_ddl_c;

pf[I_PRED_8x8_DDR] = x264_predict_8x8_ddr_c;

pf[I_PRED_8x8_VR] = x264_predict_8x8_vr_c;

pf[I_PRED_8x8_HD] = x264_predict_8x8_hd_c;

pf[I_PRED_8x8_VL] = x264_predict_8x8_vl_c;

pf[I_PRED_8x8_HU] = x264_predict_8x8_hu_c;

pf[I_PRED_8x8_DC_LEFT]= x264_predict_8x8_dc_left_c;

pf[I_PRED_8x8_DC_TOP] = x264_predict_8x8_dc_top_c;

pf[I_PRED_8x8_DC_128] = x264_predict_8x8_dc_128_c;

*predict_filter = x264_predict_8x8_filter_c;

#if HAVE_MMX

x264_predict_8x8_init_mmx( cpu, pf, predict_filter );

#endif

#if HAVE_ARMV6

x264_predict_8x8_init_arm( cpu, pf, predict_filter );

#endif

}

void x264_predict_4x4_init( int cpu, x264_predict_t pf[12] )

{

pf[I_PRED_4x4_V] = x264_predict_4x4_v_c;

pf[I_PRED_4x4_H] = x264_predict_4x4_h_c;

pf[I_PRED_4x4_DC] = x264_predict_4x4_dc_c;

pf[I_PRED_4x4_DDL] = x264_predict_4x4_ddl_c;

pf[I_PRED_4x4_DDR] = x264_predict_4x4_ddr_c;

pf[I_PRED_4x4_VR] = x264_predict_4x4_vr_c;

pf[I_PRED_4x4_HD] = x264_predict_4x4_hd_c;

pf[I_PRED_4x4_VL] = x264_predict_4x4_vl_c;

pf[I_PRED_4x4_HU] = x264_predict_4x4_hu_c;

pf[I_PRED_4x4_DC_LEFT]= x264_predict_4x4_dc_left_c;

pf[I_PRED_4x4_DC_TOP] = x264_predict_4x4_dc_top_c;

pf[I_PRED_4x4_DC_128] = x264_predict_4x4_dc_128_c;

#if HAVE_MMX

x264_predict_4x4_init_mmx( cpu, pf );

#endif

#if HAVE_ARMV6

x264_predict_4x4_init_arm( cpu, pf );

#endif

}