dsputil.c

64-bits H.264 from ffmpeg 2008 version Build in VC++ 2008 no error warning, Jesse Stone, Taiwan

{
    int i;
    for(i=0;i<8;i++) {
        pixels[0] += block[0];
        pixels[1] += block[1];
        pixels[2] += block[2];
        pixels[3] += block[3];
        pixels[4] += block[4];
        pixels[5] += block[5];
        pixels[6] += block[6];
        pixels[7] += block[7];
        pixels += line_size;
        block += 8;
    }
}

static void add_pixels4_c(uint8_t *restrict pixels, DCTELEM *block, int line_size)
{
    int i;
    for(i=0;i<4;i++) {
        pixels[0] += block[0];
        pixels[1] += block[1];
        pixels[2] += block[2];
        pixels[3] += block[3];
        pixels += line_size;
        block += 4;
    }
}

static int sum_abs_dctelem_c(DCTELEM *block)
{
    int sum=0, i;
    for(i=0; i<64; i++)
        sum+= FFABS(block[i]);
    return sum;
}

#if 0

#define PIXOP2(OPNAME, OP) \
static void OPNAME ## _pixels(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int i;\
    for(i=0; i<h; i++){\
        OP(*((uint64_t*)block), AV_RN64(pixels));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
\
static void OPNAME ## _no_rnd_pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int i;\
    for(i=0; i<h; i++){\
        const uint64_t a= AV_RN64(pixels  );\
        const uint64_t b= AV_RN64(pixels+1);\
        OP(*((uint64_t*)block), (a&b) + (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
\
static void OPNAME ## _pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int i;\
    for(i=0; i<h; i++){\
        const uint64_t a= AV_RN64(pixels  );\
        const uint64_t b= AV_RN64(pixels+1);\
        OP(*((uint64_t*)block), (a|b) - (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
\
static void OPNAME ## _no_rnd_pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int i;\
    for(i=0; i<h; i++){\
        const uint64_t a= AV_RN64(pixels          );\
        const uint64_t b= AV_RN64(pixels+line_size);\
        OP(*((uint64_t*)block), (a&b) + (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
\
static void OPNAME ## _pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int i;\
    for(i=0; i<h; i++){\
        const uint64_t a= AV_RN64(pixels          );\
        const uint64_t b= AV_RN64(pixels+line_size);\
        OP(*((uint64_t*)block), (a|b) - (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
\
static void OPNAME ## _pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
        int i;\
        const uint64_t a= AV_RN64(pixels  );\
        const uint64_t b= AV_RN64(pixels+1);\
        uint64_t l0=  (a&0x0303030303030303ULL)\
                    + (b&0x0303030303030303ULL)\
                    + 0x0202020202020202ULL;\
        uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
                   + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
        uint64_t l1,h1;\
\
        pixels+=line_size;\
        for(i=0; i<h; i+=2){\
            uint64_t a= AV_RN64(pixels  );\
            uint64_t b= AV_RN64(pixels+1);\
            l1=  (a&0x0303030303030303ULL)\
               + (b&0x0303030303030303ULL);\
            h1= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
              + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
            OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\
            pixels+=line_size;\
            block +=line_size;\
            a= AV_RN64(pixels  );\
            b= AV_RN64(pixels+1);\
            l0=  (a&0x0303030303030303ULL)\
               + (b&0x0303030303030303ULL)\
               + 0x0202020202020202ULL;\
            h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
              + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
            OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\
            pixels+=line_size;\
            block +=line_size;\
        }\
}\
\
static void OPNAME ## _no_rnd_pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
        int i;\
        const uint64_t a= AV_RN64(pixels  );\
        const uint64_t b= AV_RN64(pixels+1);\
        uint64_t l0=  (a&0x0303030303030303ULL)\
                    + (b&0x0303030303030303ULL)\
                    + 0x0101010101010101ULL;\
        uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
                   + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
        uint64_t l1,h1;\
\
        pixels+=line_size;\
        for(i=0; i<h; i+=2){\
            uint64_t a= AV_RN64(pixels  );\
            uint64_t b= AV_RN64(pixels+1);\
            l1=  (a&0x0303030303030303ULL)\
               + (b&0x0303030303030303ULL);\
            h1= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
              + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
            OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\
            pixels+=line_size;\
            block +=line_size;\
            a= AV_RN64(pixels  );\
            b= AV_RN64(pixels+1);\
            l0=  (a&0x0303030303030303ULL)\
               + (b&0x0303030303030303ULL)\
               + 0x0101010101010101ULL;\
            h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\
              + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\
            OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\
            pixels+=line_size;\
            block +=line_size;\
        }\
}\
\
CALL_2X_PIXELS(OPNAME ## _pixels16_c    , OPNAME ## _pixels_c    , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels_x2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels_y2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels_xy2_c, 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels_x2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels_y2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels_xy2_c, 8)

#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEFEFEFEFEULL)>>1) )
#else // 64 bit variant

#define PIXOP2(OPNAME, OP) \
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  )), AV_RN16(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  )), AV_RN32(pixels  ));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
static void OPNAME ## _pixels8_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    int i;\
    for(i=0; i<h; i++){\
        OP(*((uint32_t*)(block  )), AV_RN32(pixels  ));\
        OP(*((uint32_t*)(block+4)), AV_RN32(pixels+4));\
        pixels+=line_size;\
        block +=line_size;\
    }\
}\
static  void OPNAME ## _no_rnd_pixels8_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    OPNAME ## _pixels8_c(block, pixels, line_size, h);\
}\
\
static  void OPNAME ## _no_rnd_pixels8_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
                                                int src_stride1, int src_stride2, int h){\
    int i;\
    for(i=0; i<h; i++){\
        uint32_t a,b;\
        a= AV_RN32(&src1[i*src_stride1  ]);\
        b= AV_RN32(&src2[i*src_stride2  ]);\
        OP(*((uint32_t*)&dst[i*dst_stride  ]), no_rnd_avg32(a, b));\
        a= AV_RN32(&src1[i*src_stride1+4]);\
        b= AV_RN32(&src2[i*src_stride2+4]);\
        OP(*((uint32_t*)&dst[i*dst_stride+4]), no_rnd_avg32(a, b));\
    }\
}\
\
static  void OPNAME ## _pixels8_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
                                                int src_stride1, int src_stride2, int h){\
    int i;\
    for(i=0; i<h; i++){\
        uint32_t a,b;\
        a= AV_RN32(&src1[i*src_stride1  ]);\
        b= AV_RN32(&src2[i*src_stride2  ]);\
        OP(*((uint32_t*)&dst[i*dst_stride  ]), rnd_avg32(a, b));\
        a= AV_RN32(&src1[i*src_stride1+4]);\
        b= AV_RN32(&src2[i*src_stride2+4]);\
        OP(*((uint32_t*)&dst[i*dst_stride+4]), rnd_avg32(a, b));\
    }\
}\
\
static  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= AV_RN32(&src1[i*src_stride1  ]);\
        b= AV_RN32(&src2[i*src_stride2  ]);\
        OP(*((uint32_t*)&dst[i*dst_stride  ]), rnd_avg32(a, b));\
    }\
}\
\
static  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= AV_RN16(&src1[i*src_stride1  ]);\
        b= AV_RN16(&src2[i*src_stride2  ]);\
        OP(*((uint16_t*)&dst[i*dst_stride  ]), rnd_avg32(a, b));\
    }\
}\
\
static  void OPNAME ## _pixels16_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
                                                int src_stride1, int src_stride2, int h){\
    OPNAME ## _pixels8_l2(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
    OPNAME ## _pixels8_l2(dst+8, src1+8, src2+8, dst_stride, src_stride1, src_stride2, h);\
}\
\
static  void OPNAME ## _no_rnd_pixels16_l2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
                                                int src_stride1, int src_stride2, int h){\
    OPNAME ## _no_rnd_pixels8_l2(dst  , src1  , src2  , dst_stride, src_stride1, src_stride2, h);\
    OPNAME ## _no_rnd_pixels8_l2(dst+8, src1+8, src2+8, dst_stride, src_stride1, src_stride2, h);\
}\
\
static  void OPNAME ## _no_rnd_pixels8_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    OPNAME ## _no_rnd_pixels8_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
}\
\
static  void OPNAME ## _pixels8_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    OPNAME ## _pixels8_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\
}\
\
static  void OPNAME ## _no_rnd_pixels8_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    OPNAME ## _no_rnd_pixels8_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static  void OPNAME ## _pixels8_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
    OPNAME ## _pixels8_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static  void OPNAME ## _pixels8_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
                 int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
    int i;\
    for(i=0; i<h; i++){\
        uint32_t a, b, c, d, l0, l1, h0, h1;\
        a= AV_RN32(&src1[i*src_stride1]);\
        b= AV_RN32(&src2[i*src_stride2]);\
        c= AV_RN32(&src3[i*src_stride3]);\
        d= AV_RN32(&src4[i*src_stride4]);\
        l0=  (a&0x03030303UL)\
           + (b&0x03030303UL)\
           + 0x02020202UL;\
        h0= ((a&0xFCFCFCFCUL)>>2)\
          + ((b&0xFCFCFCFCUL)>>2);\
        l1=  (c&0x03030303UL)\
           + (d&0x03030303UL);\
        h1= ((c&0xFCFCFCFCUL)>>2)\
          + ((d&0xFCFCFCFCUL)>>2);\
        OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
        a= AV_RN32(&src1[i*src_stride1+4]);\
        b= AV_RN32(&src2[i*src_stride2+4]);\
        c= AV_RN32(&src3[i*src_stride3+4]);\
        d= AV_RN32(&src4[i*src_stride4+4]);\
        l0=  (a&0x03030303UL)\
           + (b&0x03030303UL)\
           + 0x02020202UL;\
        h0= ((a&0xFCFCFCFCUL)>>2)\
          + ((b&0xFCFCFCFCUL)>>2);\
        l1=  (c&0x03030303UL)\
           + (d&0x03030303UL);\
        h1= ((c&0xFCFCFCFCUL)>>2)\
          + ((d&0xFCFCFCFCUL)>>2);\
        OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
    }\
}\
\
static  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  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  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  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  void OPNAME ## _no_rnd_pixels8_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\
                 int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
    int i;\
    for(i=0; i<h; i++){\
        uint32_t a, b, c, d, l0, l1, h0, h1;\
        a= AV_RN32(&src1[i*src_stride1]);\
        b= AV_RN32(&src2[i*src_stride2]);\
        c= AV_RN32(&src3[i*src_stride3]);\
        d= AV_RN32(&src4[i*src_stride4]);\
        l0=  (a&0x03030303UL)\
           + (b&0x03030303UL)\
           + 0x01010101UL;\
        h0= ((a&0xFCFCFCFCUL)>>2)\
          + ((b&0xFCFCFCFCUL)>>2);\