predict-c.c

从服务器上下的x264编码器源码……希望对大家有帮助……这个是09年4月11日的

/*****************************************************************************
 * predict.c: h264 encoder
 *****************************************************************************
 * Copyright (C) 2003-2008 x264 project
 *
 * Authors: Laurent Aimar <[e-mail]>
 *          Loren Merritt <[e-mail]>
 *
 * 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
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * 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.
 *****************************************************************************/

#include "common/common.h"
#include "predict.h"
#include "pixel.h"

extern void predict_16x16_v_mmx( uint8_t *src );
extern void predict_16x16_h_mmxext( uint8_t *src );
extern void predict_16x16_h_ssse3( uint8_t *src );
extern void predict_16x16_dc_core_mmxext( uint8_t *src, int i_dc_left );
extern void predict_16x16_dc_left_core_mmxext( uint8_t *src, int i_dc_left );
extern void predict_16x16_dc_top_mmxext( uint8_t *src );
extern void predict_16x16_p_core_mmxext( uint8_t *src, int i00, int b, int c );
extern void predict_8x8c_p_core_mmxext( uint8_t *src, int i00, int b, int c );
extern void predict_8x8c_p_core_sse2( uint8_t *src, int i00, int b, int c );
extern void predict_8x8c_dc_core_mmxext( uint8_t *src, int s2, int s3 );
extern void predict_8x8c_dc_top_mmxext( uint8_t *src );
extern void predict_8x8c_v_mmx( uint8_t *src );
extern void predict_8x8c_h_mmxext( uint8_t *src );
extern void predict_8x8c_h_ssse3( uint8_t *src );
extern void predict_8x8_v_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_h_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hd_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hu_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_dc_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_dc_top_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_dc_left_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_ddl_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_ddr_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_ddl_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_ddr_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_vl_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_vr_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hu_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hd_sse2( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_vr_core_mmxext( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hd_ssse3( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_hu_ssse3( uint8_t *src, uint8_t edge[33] );
extern void predict_8x8_filter_mmxext   ( uint8_t *src, uint8_t edge[33], int i_neighbor, int i_filters );
extern void predict_8x8_filter_ssse3   ( uint8_t *src, uint8_t edge[33], int i_neighbor, int i_filters );
extern void predict_4x4_ddl_mmxext( uint8_t *src );
extern void predict_4x4_ddr_mmxext( uint8_t *src );
extern void predict_4x4_vl_mmxext( uint8_t *src );
extern void predict_4x4_vr_mmxext( uint8_t *src );
extern void predict_4x4_vr_ssse3( uint8_t *src );
extern void predict_4x4_hd_mmxext( uint8_t *src );
extern void predict_4x4_hd_ssse3( uint8_t *src );
extern void predict_4x4_dc_mmxext( uint8_t *src );
extern void predict_4x4_ddr_ssse3( uint8_t *src );
extern void predict_4x4_hu_mmxext( uint8_t *src );
extern void predict_16x16_dc_top_sse2( uint8_t *src );
extern void predict_16x16_dc_core_sse2( uint8_t *src, int i_dc_left );
extern void predict_16x16_dc_left_core_sse2( uint8_t *src, int i_dc_left );
extern void predict_16x16_v_sse2( uint8_t *src );
extern void predict_16x16_p_core_sse2( uint8_t *src, int i00, int b, int c );

DECLARE_ALIGNED_8( static const int8_t pb_12345678[8] ) = {1,2,3,4,5,6,7,8};
DECLARE_ALIGNED_8( static const int8_t pb_m87654321[8] ) = {-8,-7,-6,-5,-4,-3,-2,-1};
DECLARE_ALIGNED_8( static const int8_t pb_m32101234[8] ) = {-3,-2,-1,0,1,2,3,4};

#define PREDICT_P_SUM(j,i)\
    H += i * ( src[j+i - FDEC_STRIDE ]  - src[j-i - FDEC_STRIDE ] );\
    V += i * ( src[(j+i)*FDEC_STRIDE -1] - src[(j-i)*FDEC_STRIDE -1] );\

#define PREDICT_16x16_P(name)\
static void predict_16x16_p_##name( uint8_t *src )\
{\
    int a, b, c;\
    int H = 0;\
    int V = 0;\
    int i00;\
    PREDICT_P_SUM(7,1) \
    PREDICT_P_SUM(7,2) \
    PREDICT_P_SUM(7,3) \
    PREDICT_P_SUM(7,4) \
    PREDICT_P_SUM(7,5) \
    PREDICT_P_SUM(7,6) \
    PREDICT_P_SUM(7,7) \
    PREDICT_P_SUM(7,8) \
    a = 16 * ( src[15*FDEC_STRIDE -1] + src[15 - FDEC_STRIDE] );\
    b = ( 5 * H + 32 ) >> 6;\
    c = ( 5 * V + 32 ) >> 6;\
    i00 = a - b * 7 - c * 7 + 16;\
    predict_16x16_p_core_##name( src, i00, b, c );\
}

#ifndef ARCH_X86_64
PREDICT_16x16_P( mmxext )
#endif
PREDICT_16x16_P( sse2   )

#ifdef __GNUC__
static void predict_16x16_p_ssse3( uint8_t *src )
{
    int a, b, c, i00;
    int H, V;
    asm (
        "movq           %1, %%mm1 \n"
        "movq         8+%1, %%mm0 \n"
        "palignr $7, -8+%1, %%mm1 \n"
        "pmaddubsw      %2, %%mm0 \n"
        "pmaddubsw      %3, %%mm1 \n"
        "paddw       %%mm1, %%mm0 \n"
        "pshufw $14, %%mm0, %%mm1 \n"
        "paddw       %%mm1, %%mm0 \n"
        "pshufw  $1, %%mm0, %%mm1 \n"
        "paddw       %%mm1, %%mm0 \n"
        "movd        %%mm0, %0    \n"
        "movsx         %w0, %0    \n"
        :"=r"(H)
        :"m"(src[-FDEC_STRIDE]), "m"(*pb_12345678), "m"(*pb_m87654321)
    );
    V = 8 * ( src[15*FDEC_STRIDE-1] - src[-1*FDEC_STRIDE-1] )
      + 7 * ( src[14*FDEC_STRIDE-1] - src[ 0*FDEC_STRIDE-1] )
      + 6 * ( src[13*FDEC_STRIDE-1] - src[ 1*FDEC_STRIDE-1] )
      + 5 * ( src[12*FDEC_STRIDE-1] - src[ 2*FDEC_STRIDE-1] )
      + 4 * ( src[11*FDEC_STRIDE-1] - src[ 3*FDEC_STRIDE-1] )
      + 3 * ( src[10*FDEC_STRIDE-1] - src[ 4*FDEC_STRIDE-1] )
      + 2 * ( src[ 9*FDEC_STRIDE-1] - src[ 5*FDEC_STRIDE-1] )
      + 1 * ( src[ 8*FDEC_STRIDE-1] - src[ 6*FDEC_STRIDE-1] );
    a = 16 * ( src[15*FDEC_STRIDE -1] + src[15 - FDEC_STRIDE] );
    b = ( 5 * H + 32 ) >> 6;
    c = ( 5 * V + 32 ) >> 6;
    i00 = a - b * 7 - c * 7 + 16;
    predict_16x16_p_core_sse2( src, i00, b, c );
}
#endif

#define PREDICT_8x8_P(name)\
static void predict_8x8c_p_##name( uint8_t *src )\
{\
    int a, b, c;\
    int H = 0;\
    int V = 0;\
    int i00;\
    PREDICT_P_SUM(3,1)\
    PREDICT_P_SUM(3,2)\
    PREDICT_P_SUM(3,3)\
    PREDICT_P_SUM(3,4)\
    a = 16 * ( src[7*FDEC_STRIDE -1] + src[7 - FDEC_STRIDE] );\
    b = ( 17 * H + 16 ) >> 5;\
    c = ( 17 * V + 16 ) >> 5;\
    i00 = a -3*b -3*c + 16;\
    predict_8x8c_p_core_##name( src, i00, b, c );\
}

#ifndef ARCH_X86_64
PREDICT_8x8_P( mmxext )
#endif
PREDICT_8x8_P( sse2   )

#ifdef __GNUC__
static void predict_8x8c_p_ssse3( uint8_t *src )
{
    int a, b, c, i00;
    int H, V;
    asm (
        "movq           %1, %%mm0 \n"
        "pmaddubsw      %2, %%mm0 \n"
        "pshufw $14, %%mm0, %%mm1 \n"
        "paddw       %%mm1, %%mm0 \n"
        "pshufw  $1, %%mm0, %%mm1 \n"
        "paddw       %%mm1, %%mm0 \n"
        "movd        %%mm0, %0    \n"
        "movsx         %w0, %0    \n"
        :"=r"(H)
        :"m"(src[-FDEC_STRIDE]), "m"(*pb_m32101234)
    );
    V = 1 * ( src[4*FDEC_STRIDE -1] - src[ 2*FDEC_STRIDE -1] )
      + 2 * ( src[5*FDEC_STRIDE -1] - src[ 1*FDEC_STRIDE -1] )
      + 3 * ( src[6*FDEC_STRIDE -1] - src[ 0*FDEC_STRIDE -1] )
      + 4 * ( src[7*FDEC_STRIDE -1] - src[-1*FDEC_STRIDE -1] );
    H += -4 * src[-1*FDEC_STRIDE -1];
    a = 16 * ( src[7*FDEC_STRIDE -1] + src[7 - FDEC_STRIDE] );
    b = ( 17 * H + 16 ) >> 5;
    c = ( 17 * V + 16 ) >> 5;
    i00 = a -3*b -3*c + 16;
    predict_8x8c_p_core_sse2( src, i00, b, c );
}
#endif

#define PREDICT_16x16_DC(name)\
static void predict_16x16_dc_##name( uint8_t *src )\
{\
    uint32_t dc=16;\
    int i;\
    for( i = 0; i < 16; i+=2 )\
    {\
        dc += src[-1 + i * FDEC_STRIDE];\
        dc += src[-1 + (i+1) * FDEC_STRIDE];\
    }\
    predict_16x16_dc_core_##name( src, dc );\
}

PREDICT_16x16_DC( mmxext )
PREDICT_16x16_DC( sse2 )

#define PREDICT_16x16_DC_LEFT(name)\
static void predict_16x16_dc_left_##name( uint8_t *src )\
{\
    uint32_t dc=8;\
    int i;\
    for( i = 0; i < 16; i+=2 )\
    {\
        dc += src[-1 + i * FDEC_STRIDE];\
        dc += src[-1 + (i+1) * FDEC_STRIDE];\
    }\
    predict_16x16_dc_left_core_##name( src, dc>>4 );\
}

PREDICT_16x16_DC_LEFT( mmxext )
PREDICT_16x16_DC_LEFT( sse2 )

static void predict_8x8c_dc_mmxext( uint8_t *src )
{
    int s2 = 4
       + src[-1 + 0*FDEC_STRIDE]
       + src[-1 + 1*FDEC_STRIDE]