X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=common%2Fpredict.c;h=158136cc528968ac7dcde7e45a1cbc1f507cda18;hb=ce6c94c0bef3350e9546302aae5909404b056fdb;hp=2c4e83b5cc2002d42c76e3bdc23c0f8f9e9d23c5;hpb=bdbd4fe7709e129f90cf3d7d59b500e915c6b187;p=x264 diff --git a/common/predict.c b/common/predict.c index 2c4e83b5..158136cc 100644 --- a/common/predict.c +++ b/common/predict.c @@ -1,11 +1,12 @@ /***************************************************************************** - * predict.c: h264 encoder + * predict.c: intra prediction ***************************************************************************** - * Copyright (C) 2003-2008 x264 project + * Copyright (C) 2003-2014 x264 project * * Authors: Laurent Aimar * Loren Merritt * Fiona Glaser + * Henrik Gramner * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -20,6 +21,9 @@ * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. + * + * This program is also available under a commercial proprietary license. + * For more information, contact us at licensing@x264.com. *****************************************************************************/ /* predict4x4 are inspired from ffmpeg h264 decoder */ @@ -27,139 +31,121 @@ #include "common.h" -#ifdef _MSC_VER -#undef HAVE_MMX /* not finished now */ -#endif -#ifdef HAVE_MMX +#if HAVE_MMX # include "x86/predict.h" #endif -#ifdef ARCH_PPC +#if ARCH_PPC # include "ppc/predict.h" #endif +#if ARCH_ARM +# include "arm/predict.h" +#endif +#if ARCH_AARCH64 +# include "aarch64/predict.h" +#endif /**************************************************************************** * 16x16 prediction for intra luma block ****************************************************************************/ -#define PREDICT_16x16_DC(v) \ - for( i = 0; i < 16; i++ )\ +#define PREDICT_16x16_DC(v)\ + for( int i = 0; i < 16; i++ )\ {\ - uint32_t *p = (uint32_t*)src;\ - *p++ = v;\ - *p++ = v;\ - *p++ = v;\ - *p++ = v;\ + MPIXEL_X4( src+ 0 ) = v;\ + MPIXEL_X4( src+ 4 ) = v;\ + MPIXEL_X4( src+ 8 ) = v;\ + MPIXEL_X4( src+12 ) = v;\ src += FDEC_STRIDE;\ } -static void predict_16x16_dc( uint8_t *src ) +void x264_predict_16x16_dc_c( pixel *src ) { - uint32_t dc = 0; - int i; + int dc = 0; - for( i = 0; i < 16; i++ ) + for( int i = 0; i < 16; i++ ) { dc += src[-1 + i * FDEC_STRIDE]; dc += src[i - FDEC_STRIDE]; } - dc = (( dc + 16 ) >> 5) * 0x01010101; + pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 16 ) >> 5 ); - PREDICT_16x16_DC(dc); + PREDICT_16x16_DC( dcsplat ); } -static void predict_16x16_dc_left( uint8_t *src ) +static void x264_predict_16x16_dc_left_c( pixel *src ) { - uint32_t dc = 0; - int i; + int dc = 0; - for( i = 0; i < 16; i++ ) - { + for( int i = 0; i < 16; i++ ) dc += src[-1 + i * FDEC_STRIDE]; - } - dc = (( dc + 8 ) >> 4) * 0x01010101; + pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 ); - PREDICT_16x16_DC(dc); + PREDICT_16x16_DC( dcsplat ); } -static void predict_16x16_dc_top( uint8_t *src ) +static void x264_predict_16x16_dc_top_c( pixel *src ) { - uint32_t dc = 0; - int i; + int dc = 0; - for( i = 0; i < 16; i++ ) - { + for( int i = 0; i < 16; i++ ) dc += src[i - FDEC_STRIDE]; - } - dc = (( dc + 8 ) >> 4) * 0x01010101; + pixel4 dcsplat = PIXEL_SPLAT_X4( ( dc + 8 ) >> 4 ); - PREDICT_16x16_DC(dc); + PREDICT_16x16_DC( dcsplat ); } -static void predict_16x16_dc_128( uint8_t *src ) +static void x264_predict_16x16_dc_128_c( pixel *src ) { - int i; - PREDICT_16x16_DC(0x80808080); + PREDICT_16x16_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) ); } -static void predict_16x16_h( uint8_t *src ) +void x264_predict_16x16_h_c( pixel *src ) { - int i; - - for( i = 0; i < 16; i++ ) + for( int i = 0; i < 16; i++ ) { - const uint32_t v = 0x01010101 * src[-1]; - uint32_t *p = (uint32_t*)src; - - *p++ = v; - *p++ = v; - *p++ = v; - *p++ = v; - + 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; - } } -static void predict_16x16_v( uint8_t *src ) +void x264_predict_16x16_v_c( pixel *src ) { - uint32_t v0 = *(uint32_t*)&src[ 0-FDEC_STRIDE]; - uint32_t v1 = *(uint32_t*)&src[ 4-FDEC_STRIDE]; - uint32_t v2 = *(uint32_t*)&src[ 8-FDEC_STRIDE]; - uint32_t v3 = *(uint32_t*)&src[12-FDEC_STRIDE]; - int i; + 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( i = 0; i < 16; i++ ) + for( int i = 0; i < 16; i++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = v0; - *p++ = v1; - *p++ = v2; - *p++ = v3; + MPIXEL_X4( src+ 0 ) = v0; + MPIXEL_X4( src+ 4 ) = v1; + MPIXEL_X4( src+ 8 ) = v2; + MPIXEL_X4( src+12 ) = v3; src += FDEC_STRIDE; } } -static void predict_16x16_p( uint8_t *src ) +void x264_predict_16x16_p_c( pixel *src ) { - int x, y, i; - int a, b, c; - int H = 0; - int V = 0; - int i00; + int H = 0, V = 0; /* calculate H and V */ - for( i = 0; i <= 7; i++ ) + 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 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] ); - b = ( 5 * H + 32 ) >> 6; - c = ( 5 * V + 32 ) >> 6; + int a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] ); + int b = ( 5 * H + 32 ) >> 6; + int c = ( 5 * V + 32 ) >> 6; - i00 = a - b * 7 - c * 7 + 16; + int i00 = a - b * 7 - c * 7 + 16; - for( y = 0; y < 16; y++ ) + for( int y = 0; y < 16; y++ ) { int pix = i00; - for( x = 0; x < 16; x++ ) + for( int x = 0; x < 16; x++ ) { - src[x] = x264_clip_uint8( pix>>5 ); + src[x] = x264_clip_pixel( pix>>5 ); pix += b; } src += FDEC_STRIDE; @@ -169,84 +155,73 @@ static void predict_16x16_p( uint8_t *src ) /**************************************************************************** - * 8x8 prediction for intra chroma block + * 8x8 prediction for intra chroma block (4:2:0) ****************************************************************************/ -static void predict_8x8c_dc_128( uint8_t *src ) +static void x264_predict_8x8c_dc_128_c( pixel *src ) { - int y; - - for( y = 0; y < 8; y++ ) + for( int y = 0; y < 8; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = 0x80808080; - *p++ = 0x80808080; + 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 predict_8x8c_dc_left( uint8_t *src ) +static void x264_predict_8x8c_dc_left_c( pixel *src ) { - int y; - uint32_t dc0 = 0, dc1 = 0; + int dc0 = 0, dc1 = 0; - for( y = 0; y < 4; y++ ) + for( int y = 0; y < 4; y++ ) { dc0 += src[y * FDEC_STRIDE - 1]; dc1 += src[(y+4) * FDEC_STRIDE - 1]; } - dc0 = (( dc0 + 2 ) >> 2)*0x01010101; - dc1 = (( dc1 + 2 ) >> 2)*0x01010101; + pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 ); + pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 ); - for( y = 0; y < 4; y++ ) + for( int y = 0; y < 4; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = dc0; - *p++ = dc0; + MPIXEL_X4( src+0 ) = dc0splat; + MPIXEL_X4( src+4 ) = dc0splat; src += FDEC_STRIDE; } - for( y = 0; y < 4; y++ ) + for( int y = 0; y < 4; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = dc1; - *p++ = dc1; + MPIXEL_X4( src+0 ) = dc1splat; + MPIXEL_X4( src+4 ) = dc1splat; src += FDEC_STRIDE; } } -static void predict_8x8c_dc_top( uint8_t *src ) +static void x264_predict_8x8c_dc_top_c( pixel *src ) { - int y, x; - uint32_t dc0 = 0, dc1 = 0; + int dc0 = 0, dc1 = 0; - for( x = 0; x < 4; x++ ) + for( int x = 0; x < 4; x++ ) { dc0 += src[x - FDEC_STRIDE]; dc1 += src[x + 4 - FDEC_STRIDE]; } - dc0 = (( dc0 + 2 ) >> 2)*0x01010101; - dc1 = (( dc1 + 2 ) >> 2)*0x01010101; + pixel4 dc0splat = PIXEL_SPLAT_X4( ( dc0 + 2 ) >> 2 ); + pixel4 dc1splat = PIXEL_SPLAT_X4( ( dc1 + 2 ) >> 2 ); - for( y = 0; y < 8; y++ ) + for( int y = 0; y < 8; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = dc0; - *p++ = dc1; + MPIXEL_X4( src+0 ) = dc0splat; + MPIXEL_X4( src+4 ) = dc1splat; src += FDEC_STRIDE; } } -static void predict_8x8c_dc( uint8_t *src ) +void x264_predict_8x8c_dc_c( pixel *src ) { - int y; int s0 = 0, s1 = 0, s2 = 0, s3 = 0; - uint32_t dc0, dc1, dc2, dc3; - int i; /* s0 s1 s2 s3 */ - for( i = 0; i < 4; i++ ) + for( int i = 0; i < 4; i++ ) { s0 += src[i - FDEC_STRIDE]; s1 += src[i + 4 - FDEC_STRIDE]; @@ -257,80 +232,229 @@ static void predict_8x8c_dc( uint8_t *src ) dc0 dc1 dc2 dc3 */ - dc0 = (( s0 + s2 + 4 ) >> 3)*0x01010101; - dc1 = (( s1 + 2 ) >> 2)*0x01010101; - dc2 = (( s3 + 2 ) >> 2)*0x01010101; - dc3 = (( s1 + s3 + 4 ) >> 3)*0x01010101; + 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 ); - for( y = 0; y < 4; y++ ) + for( int y = 0; y < 4; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = dc0; - *p++ = dc1; + MPIXEL_X4( src+0 ) = dc0; + MPIXEL_X4( src+4 ) = dc1; src += FDEC_STRIDE; } - for( y = 0; y < 4; y++ ) + for( int y = 0; y < 4; y++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = dc2; - *p++ = dc3; + MPIXEL_X4( src+0 ) = dc2; + MPIXEL_X4( src+4 ) = dc3; src += FDEC_STRIDE; } } -static void predict_8x8c_h( uint8_t *src ) +void x264_predict_8x8c_h_c( pixel *src ) { - int i; - - for( i = 0; i < 8; i++ ) + for( int i = 0; i < 8; i++ ) { - uint32_t v = 0x01010101 * src[-1]; - uint32_t *p = (uint32_t*)src; - *p++ = v; - *p++ = v; + pixel4 v = PIXEL_SPLAT_X4( src[-1] ); + MPIXEL_X4( src+0 ) = v; + MPIXEL_X4( src+4 ) = v; src += FDEC_STRIDE; } } -static void predict_8x8c_v( uint8_t *src ) +void x264_predict_8x8c_v_c( pixel *src ) { - uint32_t v0 = *(uint32_t*)&src[0-FDEC_STRIDE]; - uint32_t v1 = *(uint32_t*)&src[4-FDEC_STRIDE]; - int i; + pixel4 v0 = MPIXEL_X4( src+0-FDEC_STRIDE ); + pixel4 v1 = MPIXEL_X4( src+4-FDEC_STRIDE ); - for( i = 0; i < 8; i++ ) + for( int i = 0; i < 8; i++ ) { - uint32_t *p = (uint32_t*)src; - *p++ = v0; - *p++ = v1; + MPIXEL_X4( src+0 ) = v0; + MPIXEL_X4( src+4 ) = v1; src += FDEC_STRIDE; } } -static void predict_8x8c_p( uint8_t *src ) +void x264_predict_8x8c_p_c( pixel *src ) { - int i; - int x,y; - int a, b, c; - int H = 0; - int V = 0; - int i00; + int H = 0, V = 0; - for( i = 0; i < 4; i++ ) + 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 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] ); - b = ( 17 * H + 16 ) >> 5; - c = ( 17 * V + 16 ) >> 5; - i00 = a -3*b -3*c + 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 ) +{ + 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 ) +{ + 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; + + 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 + */ + 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 + */ + 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 ) +{ + 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 ); + + 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; + + 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] ); - for( y = 0; y < 8; y++ ) + 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( x = 0; x < 8; x++ ) + for( int x = 0; x < 8; x++ ) { - src[x] = x264_clip_uint8( pix>>5 ); + src[x] = x264_clip_pixel( pix>>5 ); pix += b; } src += FDEC_STRIDE; @@ -343,65 +467,65 @@ static void predict_8x8c_p( uint8_t *src ) ****************************************************************************/ #define SRC(x,y) src[(x)+(y)*FDEC_STRIDE] -#define SRC32(x,y) *(uint32_t*)&SRC(x,y) +#define SRC_X4(x,y) MPIXEL_X4( &SRC(x,y) ) #define PREDICT_4x4_DC(v)\ - SRC32(0,0) = SRC32(0,1) = SRC32(0,2) = SRC32(0,3) = v; + SRC_X4(0,0) = SRC_X4(0,1) = SRC_X4(0,2) = SRC_X4(0,3) = v; -static void predict_4x4_dc_128( uint8_t *src ) +static void x264_predict_4x4_dc_128_c( pixel *src ) { - PREDICT_4x4_DC(0x80808080); + PREDICT_4x4_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) ); } -static void predict_4x4_dc_left( uint8_t *src ) +static void x264_predict_4x4_dc_left_c( pixel *src ) { - uint32_t dc = ((SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2) * 0x01010101; - PREDICT_4x4_DC(dc); + pixel4 dc = PIXEL_SPLAT_X4( (SRC(-1,0) + SRC(-1,1) + SRC(-1,2) + SRC(-1,3) + 2) >> 2 ); + PREDICT_4x4_DC( dc ); } -static void predict_4x4_dc_top( uint8_t *src ) +static void x264_predict_4x4_dc_top_c( pixel *src ) { - uint32_t dc = ((SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2) * 0x01010101; - PREDICT_4x4_DC(dc); + pixel4 dc = PIXEL_SPLAT_X4( (SRC(0,-1) + SRC(1,-1) + SRC(2,-1) + SRC(3,-1) + 2) >> 2 ); + PREDICT_4x4_DC( dc ); } -static void predict_4x4_dc( uint8_t *src ) +void x264_predict_4x4_dc_c( pixel *src ) { - uint32_t dc = ((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) * 0x01010101; - PREDICT_4x4_DC(dc); + 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 ); } -static void predict_4x4_h( uint8_t *src ) +void x264_predict_4x4_h_c( pixel *src ) { - SRC32(0,0) = SRC(-1,0) * 0x01010101; - SRC32(0,1) = SRC(-1,1) * 0x01010101; - SRC32(0,2) = SRC(-1,2) * 0x01010101; - SRC32(0,3) = SRC(-1,3) * 0x01010101; + 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) ); } -static void predict_4x4_v( uint8_t *src ) +void x264_predict_4x4_v_c( pixel *src ) { - PREDICT_4x4_DC(SRC32(0,-1)); + PREDICT_4x4_DC(SRC_X4(0,-1)); } #define PREDICT_4x4_LOAD_LEFT\ - const int l0 = SRC(-1,0);\ - const int l1 = SRC(-1,1);\ - const int l2 = SRC(-1,2);\ - UNUSED const int l3 = SRC(-1,3); + int l0 = SRC(-1,0);\ + int l1 = SRC(-1,1);\ + int l2 = SRC(-1,2);\ + UNUSED int l3 = SRC(-1,3); #define PREDICT_4x4_LOAD_TOP\ - const int t0 = SRC(0,-1);\ - const int t1 = SRC(1,-1);\ - const int t2 = SRC(2,-1);\ - UNUSED const int t3 = SRC(3,-1); + int t0 = SRC(0,-1);\ + int t1 = SRC(1,-1);\ + int t2 = SRC(2,-1);\ + UNUSED int t3 = SRC(3,-1); #define PREDICT_4x4_LOAD_TOP_RIGHT\ - const int t4 = SRC(4,-1);\ - const int t5 = SRC(5,-1);\ - const int t6 = SRC(6,-1);\ - UNUSED const int t7 = SRC(7,-1); + int t4 = SRC(4,-1);\ + int t5 = SRC(5,-1);\ + int t6 = SRC(6,-1);\ + UNUSED int t7 = SRC(7,-1); #define F1(a,b) (((a)+(b)+1)>>1) #define F2(a,b,c) (((a)+2*(b)+(c)+2)>>2) -static void predict_4x4_ddl( uint8_t *src ) +static void x264_predict_4x4_ddl_c( pixel *src ) { PREDICT_4x4_LOAD_TOP PREDICT_4x4_LOAD_TOP_RIGHT @@ -413,9 +537,9 @@ static void predict_4x4_ddl( uint8_t *src ) SRC(3,2)=SRC(2,3)= F2(t5,t6,t7); SRC(3,3)= F2(t6,t7,t7); } -static void predict_4x4_ddr( uint8_t *src ) +static void x264_predict_4x4_ddr_c( pixel *src ) { - const int lt = SRC(-1,-1); + int lt = SRC(-1,-1); PREDICT_4x4_LOAD_LEFT PREDICT_4x4_LOAD_TOP SRC(3,0)= F2(t3,t2,t1); @@ -427,9 +551,9 @@ static void predict_4x4_ddr( uint8_t *src ) SRC(0,3)= F2(l1,l2,l3); } -static void predict_4x4_vr( uint8_t *src ) +static void x264_predict_4x4_vr_c( pixel *src ) { - const int lt = SRC(-1,-1); + int lt = SRC(-1,-1); PREDICT_4x4_LOAD_LEFT PREDICT_4x4_LOAD_TOP SRC(0,3)= F2(l2,l1,l0); @@ -444,9 +568,9 @@ static void predict_4x4_vr( uint8_t *src ) SRC(3,0)= F1(t2,t3); } -static void predict_4x4_hd( uint8_t *src ) +static void x264_predict_4x4_hd_c( pixel *src ) { - const int lt= SRC(-1,-1); + int lt= SRC(-1,-1); PREDICT_4x4_LOAD_LEFT PREDICT_4x4_LOAD_TOP SRC(0,3)= F1(l2,l3); @@ -461,7 +585,7 @@ static void predict_4x4_hd( uint8_t *src ) SRC(3,0)= F2(t2,t1,t0); } -static void predict_4x4_vl( uint8_t *src ) +static void x264_predict_4x4_vl_c( pixel *src ) { PREDICT_4x4_LOAD_TOP PREDICT_4x4_LOAD_TOP_RIGHT @@ -477,7 +601,7 @@ static void predict_4x4_vl( uint8_t *src ) SRC(3,3)= F2(t4,t5,t6); } -static void predict_4x4_hu( uint8_t *src ) +static void x264_predict_4x4_hu_c( pixel *src ) { PREDICT_4x4_LOAD_LEFT SRC(0,0)= F1(l0,l1); @@ -499,7 +623,7 @@ static void predict_4x4_hu( uint8_t *src ) #define PT(x) \ edge[16+x] = F2(SRC(x-1,-1), SRC(x,-1), SRC(x+1,-1)); -void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, int i_filters ) +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 @@ -509,10 +633,11 @@ void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, in int have_lt = i_neighbor & MB_TOPLEFT; if( i_filters & MB_LEFT ) { - edge[15] = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2; + 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; + + 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; } @@ -520,10 +645,10 @@ void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, in { 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; + + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) - edge[23] = ((have_tr ? SRC(8,-1) : SRC(7,-1)) - + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2; + edge[23] = (SRC(6,-1) + 2*SRC(7,-1) + + (have_tr ? SRC(8,-1) : SRC(7,-1)) + 2) >> 2; if( i_filters & MB_TOPRIGHT ) { @@ -535,7 +660,8 @@ void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, in } else { - *(uint64_t*)(edge+24) = SRC(7,-1) * 0x0101010101010101ULL; + 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); } } @@ -546,11 +672,11 @@ void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, in #undef PT #define PL(y) \ - UNUSED const int l##y = edge[14-y]; + UNUSED int l##y = edge[14-y]; #define PT(x) \ - UNUSED const int t##x = edge[16+x]; + UNUSED int t##x = edge[16+x]; #define PREDICT_8x8_LOAD_TOPLEFT \ - const int lt = edge[15]; + 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 \ @@ -559,54 +685,54 @@ void x264_predict_8x8_filter( uint8_t *src, uint8_t edge[33], int i_neighbor, in PT(8) PT(9) PT(10) PT(11) PT(12) PT(13) PT(14) PT(15) #define PREDICT_8x8_DC(v) \ - int y; \ - for( y = 0; y < 8; y++ ) { \ - ((uint32_t*)src)[0] = \ - ((uint32_t*)src)[1] = v; \ + for( int y = 0; y < 8; y++ ) { \ + MPIXEL_X4( src+0 ) = v; \ + MPIXEL_X4( src+4 ) = v; \ src += FDEC_STRIDE; \ } -/* SIMD is much faster than C for all of these except HU and HD. */ -static void predict_8x8_dc_128( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_dc_128_c( pixel *src, pixel edge[36] ) { - PREDICT_8x8_DC(0x80808080); + PREDICT_8x8_DC( PIXEL_SPLAT_X4( 1 << (BIT_DEPTH-1) ) ); } -static void predict_8x8_dc_left( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_dc_left_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_LEFT - const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101; - PREDICT_8x8_DC(dc); + pixel4 dc = PIXEL_SPLAT_X4( (l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3 ); + PREDICT_8x8_DC( dc ); } -static void predict_8x8_dc_top( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_dc_top_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_TOP - const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101; - PREDICT_8x8_DC(dc); + pixel4 dc = PIXEL_SPLAT_X4( (t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3 ); + PREDICT_8x8_DC( dc ); } -static void predict_8x8_dc( uint8_t *src, uint8_t edge[33] ) +void x264_predict_8x8_dc_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_LEFT PREDICT_8x8_LOAD_TOP - const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7 - +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101; - PREDICT_8x8_DC(dc); + 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 ); } -static void predict_8x8_h( uint8_t *src, uint8_t edge[33] ) +void x264_predict_8x8_h_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_LEFT -#define ROW(y) ((uint32_t*)(src+y*FDEC_STRIDE))[0] =\ - ((uint32_t*)(src+y*FDEC_STRIDE))[1] = 0x01010101U * l##y +#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 } -static void predict_8x8_v( uint8_t *src, uint8_t edge[33] ) +void x264_predict_8x8_v_c( pixel *src, pixel edge[36] ) { - const uint64_t top = *(uint64_t*)(edge+16); - int y; - for( y = 0; y < 8; y++ ) - *(uint64_t*)(src+y*FDEC_STRIDE) = top; + 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 predict_8x8_ddl( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_ddl_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_TOP PREDICT_8x8_LOAD_TOPRIGHT @@ -626,7 +752,7 @@ static void predict_8x8_ddl( uint8_t *src, uint8_t edge[33] ) SRC(6,7)=SRC(7,6)= F2(t13,t14,t15); SRC(7,7)= F2(t14,t15,t15); } -static void predict_8x8_ddr( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_ddr_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_TOP PREDICT_8x8_LOAD_LEFT @@ -646,9 +772,9 @@ static void predict_8x8_ddr( uint8_t *src, uint8_t edge[33] ) 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 predict_8x8_vr( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_vr_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_TOP PREDICT_8x8_LOAD_LEFT @@ -676,34 +802,34 @@ static void predict_8x8_vr( uint8_t *src, uint8_t edge[33] ) SRC(7,1)= F2(t5,t6,t7); SRC(7,0)= F1(t6,t7); } -static void predict_8x8_hd( uint8_t *src, uint8_t edge[33] ) +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 = pack8to16(F1(l6,l7), F2(l5,l6,l7)); - int p2 = pack8to16(F1(l5,l6), F2(l4,l5,l6)); - int p3 = pack8to16(F1(l4,l5), F2(l3,l4,l5)); - int p4 = pack8to16(F1(l3,l4), F2(l2,l3,l4)); - int p5 = pack8to16(F1(l2,l3), F2(l1,l2,l3)); - int p6 = pack8to16(F1(l1,l2), F2(l0,l1,l2)); - int p7 = pack8to16(F1(l0,l1), F2(lt,l0,l1)); - int p8 = pack8to16(F1(lt,l0), F2(l0,lt,t0)); - int p9 = pack8to16(F2(t1,t0,lt), F2(t2,t1,t0)); - int p10 = pack8to16(F2(t3,t2,t1), F2(t4,t3,t2)); - int p11 = pack8to16(F2(t5,t4,t3), F2(t6,t5,t4)); - SRC32(0,7)= pack16to32(p1,p2); - SRC32(0,6)= pack16to32(p2,p3); - SRC32(4,7)=SRC32(0,5)= pack16to32(p3,p4); - SRC32(4,6)=SRC32(0,4)= pack16to32(p4,p5); - SRC32(4,5)=SRC32(0,3)= pack16to32(p5,p6); - SRC32(4,4)=SRC32(0,2)= pack16to32(p6,p7); - SRC32(4,3)=SRC32(0,1)= pack16to32(p7,p8); - SRC32(4,2)=SRC32(0,0)= pack16to32(p8,p9); - SRC32(4,1)= pack16to32(p9,p10); - SRC32(4,0)= pack16to32(p10,p11); -} -static void predict_8x8_vl( uint8_t *src, uint8_t edge[33] ) + 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 @@ -730,25 +856,25 @@ static void predict_8x8_vl( uint8_t *src, uint8_t edge[33] ) SRC(7,6)= F1(t10,t11); SRC(7,7)= F2(t10,t11,t12); } -static void predict_8x8_hu( uint8_t *src, uint8_t edge[33] ) +static void x264_predict_8x8_hu_c( pixel *src, pixel edge[36] ) { PREDICT_8x8_LOAD_LEFT - int p1 = pack8to16(F1(l0,l1), F2(l0,l1,l2)); - int p2 = pack8to16(F1(l1,l2), F2(l1,l2,l3)); - int p3 = pack8to16(F1(l2,l3), F2(l2,l3,l4)); - int p4 = pack8to16(F1(l3,l4), F2(l3,l4,l5)); - int p5 = pack8to16(F1(l4,l5), F2(l4,l5,l6)); - int p6 = pack8to16(F1(l5,l6), F2(l5,l6,l7)); - int p7 = pack8to16(F1(l6,l7), F2(l6,l7,l7)); - int p8 = pack8to16(l7,l7); - SRC32(0,0)= pack16to32(p1,p2); - SRC32(0,1)= pack16to32(p2,p3); - SRC32(4,0)=SRC32(0,2)= pack16to32(p3,p4); - SRC32(4,1)=SRC32(0,3)= pack16to32(p4,p5); - SRC32(4,2)=SRC32(0,4)= pack16to32(p5,p6); - SRC32(4,3)=SRC32(0,5)= pack16to32(p6,p7); - SRC32(4,4)=SRC32(0,6)= pack16to32(p7,p8); - SRC32(4,5)=SRC32(4,6)= SRC32(0,7) = SRC32(4,7) = pack16to32(p8,p8); + 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); } /**************************************************************************** @@ -756,78 +882,133 @@ static void predict_8x8_hu( uint8_t *src, uint8_t edge[33] ) ****************************************************************************/ void x264_predict_16x16_init( int cpu, x264_predict_t pf[7] ) { - pf[I_PRED_16x16_V ] = predict_16x16_v; - pf[I_PRED_16x16_H ] = predict_16x16_h; - pf[I_PRED_16x16_DC] = predict_16x16_dc; - pf[I_PRED_16x16_P ] = predict_16x16_p; - pf[I_PRED_16x16_DC_LEFT]= predict_16x16_dc_left; - pf[I_PRED_16x16_DC_TOP ]= predict_16x16_dc_top; - pf[I_PRED_16x16_DC_128 ]= predict_16x16_dc_128; + 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; -#ifdef HAVE_MMX +#if HAVE_MMX x264_predict_16x16_init_mmx( cpu, pf ); #endif -#ifdef ARCH_PPC +#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 + +#if ARCH_AARCH64 + x264_predict_16x16_init_aarch64( cpu, pf ); #endif } void x264_predict_8x8c_init( int cpu, x264_predict_t pf[7] ) { - pf[I_PRED_CHROMA_V ] = predict_8x8c_v; - pf[I_PRED_CHROMA_H ] = predict_8x8c_h; - pf[I_PRED_CHROMA_DC] = predict_8x8c_dc; - pf[I_PRED_CHROMA_P ] = predict_8x8c_p; - pf[I_PRED_CHROMA_DC_LEFT]= predict_8x8c_dc_left; - pf[I_PRED_CHROMA_DC_TOP ]= predict_8x8c_dc_top; - pf[I_PRED_CHROMA_DC_128 ]= predict_8x8c_dc_128; + 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; -#ifdef HAVE_MMX +#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 + +#if ARCH_AARCH64 + x264_predict_8x8c_init_aarch64( cpu, pf ); +#endif } -void x264_predict_8x8_init( int cpu, x264_predict8x8_t pf[12] ) +void x264_predict_8x16c_init( int cpu, x264_predict_t pf[7] ) { - pf[I_PRED_8x8_V] = predict_8x8_v; - pf[I_PRED_8x8_H] = predict_8x8_h; - pf[I_PRED_8x8_DC] = predict_8x8_dc; - pf[I_PRED_8x8_DDL] = predict_8x8_ddl; - pf[I_PRED_8x8_DDR] = predict_8x8_ddr; - pf[I_PRED_8x8_VR] = predict_8x8_vr; - pf[I_PRED_8x8_HD] = predict_8x8_hd; - pf[I_PRED_8x8_VL] = predict_8x8_vl; - pf[I_PRED_8x8_HU] = predict_8x8_hu; - pf[I_PRED_8x8_DC_LEFT]= predict_8x8_dc_left; - pf[I_PRED_8x8_DC_TOP] = predict_8x8_dc_top; - pf[I_PRED_8x8_DC_128] = predict_8x8_dc_128; + 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 + +#if ARCH_AARCH64 + x264_predict_8x16c_init_aarch64( cpu, pf ); +#endif +} -#ifdef HAVE_MMX - x264_predict_8x8_init_mmx( cpu, pf ); +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 + +#if ARCH_AARCH64 + x264_predict_8x8_init_aarch64( cpu, pf, predict_filter ); #endif } void x264_predict_4x4_init( int cpu, x264_predict_t pf[12] ) { - pf[I_PRED_4x4_V] = predict_4x4_v; - pf[I_PRED_4x4_H] = predict_4x4_h; - pf[I_PRED_4x4_DC] = predict_4x4_dc; - pf[I_PRED_4x4_DDL] = predict_4x4_ddl; - pf[I_PRED_4x4_DDR] = predict_4x4_ddr; - pf[I_PRED_4x4_VR] = predict_4x4_vr; - pf[I_PRED_4x4_HD] = predict_4x4_hd; - pf[I_PRED_4x4_VL] = predict_4x4_vl; - pf[I_PRED_4x4_HU] = predict_4x4_hu; - pf[I_PRED_4x4_DC_LEFT]= predict_4x4_dc_left; - pf[I_PRED_4x4_DC_TOP] = predict_4x4_dc_top; - pf[I_PRED_4x4_DC_128] = predict_4x4_dc_128; - -#ifdef HAVE_MMX + 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 + +#if ARCH_AARCH64 + x264_predict_4x4_init_aarch64( cpu, pf ); +#endif }