h264idct.c

君正早期ucos系统(只有早期的才不没有打包成库),MPLAYER,文件系统,图片解码,浏览,电子书,录音,想学ucos,识货的人就下吧 russblock fmradio explore set

        src[i][6] = b2 - b5;
        src[i][2] = b4 + b3;
        src[i][5] = b4 - b3;
        src[i][3] = b6 + b1;
        src[i][4] = b6 - b1;
    }
    for( i = 0; i < 8; i++ )
    {
        const int a0 =  src[0][i] + src[4][i];
        const int a2 =  src[0][i] - src[4][i];
        const int a4 = (src[2][i]>>1) - src[6][i];
        const int a6 = (src[6][i]>>1) + src[2][i];

        const int b0 = a0 + a6;
        const int b2 = a2 + a4;
        const int b4 = a2 - a4;
        const int b6 = a0 - a6;

        const int a1 = -src[3][i] + src[5][i] - src[7][i] - (src[7][i]>>1);
        const int a3 =  src[1][i] + src[7][i] - src[3][i] - (src[3][i]>>1);
        const int a5 = -src[1][i] + src[7][i] + src[5][i] + (src[5][i]>>1);
        const int a7 =  src[3][i] + src[5][i] + src[1][i] + (src[1][i]>>1);

        const int b1 = (a7>>2) + a1;
        const int b3 =  a3 + (a5>>2);
        const int b5 = (a3>>2) - a5;
        const int b7 =  a7 - (a1>>2);

        dst[i + 0*stride] = cm[ dst[i + 0*stride] + ((b0 + b7) >> 6) ];
        dst[i + 1*stride] = cm[ dst[i + 1*stride] + ((b2 + b5) >> 6) ];
        dst[i + 2*stride] = cm[ dst[i + 2*stride] + ((b4 + b3) >> 6) ];
        dst[i + 3*stride] = cm[ dst[i + 3*stride] + ((b6 + b1) >> 6) ];
        dst[i + 4*stride] = cm[ dst[i + 4*stride] + ((b6 - b1) >> 6) ];
        dst[i + 5*stride] = cm[ dst[i + 5*stride] + ((b4 - b3) >> 6) ];
        dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b2 - b5) >> 6) ];
        dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b7) >> 6) ];
    }
}

#ifdef JZ4740_MXU_OPT
// assumes all AC coefs are 0
void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
    int dc = (block[0] + 32) >> 6;
/*
    int i, j;
    uint8_t *cm = cropTbl + MAX_NEG_CROP;
    for( j = 0; j < 4; j++ )
    {
        for( i = 0; i < 4; i++ )
            dst[i] = cm[ dst[i] + dc ];
        dst += stride;
    }
*/
    S32I2M(xr1, dc);
    S32SFL(xr0,xr1,xr1,xr2, ptn3);    // Xr2 (dc, dc)
    S32SFL(xr12,xr2,xr2,xr13, ptn3);  // Xr12 (dc, dc) Xr13 (dc, dc)

//Loop 0
//i = 0 - 3
    S32LDD(xr1, dst, 0);              // Xr1 (b3, b2, b1, b0)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//Loop 1
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//Loop 2
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//Loop 3
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);
}
#else
// assumes all AC coefs are 0
void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
    int i, j;
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
    int dc = (block[0] + 32) >> 6;
    for( j = 0; j < 4; j++ )
    {
        for( i = 0; i < 4; i++ )
            dst[i] = cm[ dst[i] + dc ];
        dst += stride;
    }
}
#endif

#ifdef JZ4740_MXU_OPT
void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
    int dc = (block[0] + 32) >> 6;
/*
    int i, j;
    uint8_t *cm = cropTbl + MAX_NEG_CROP;
    for( j = 0; j < 8; j++ )
    {
        for( i = 0; i < 8; i++ )
            dst[i] = cm[ dst[i] + dc ];
        dst += stride;
    }
*/

    S32I2M(xr1, dc);
    S32SFL(xr0,xr1,xr1,xr2, ptn3);    // Xr2 (dc, dc)
    S32SFL(xr12,xr2,xr2,xr13, ptn3);  // Xr12 (dc, dc) Xr13 (dc, dc)

//Loop 0
//i = 0 - 3
    S32LDD(xr1, dst, 0);              // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 1
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 2
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 3
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 4
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 5
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 6
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);

//Loop 7
//i = 0 - 3
    S32LDIV(xr1, dst, stride, 0);     // Xr1 (b3, b2, b1, b0)
    S32LDD(xr3, dst, 4);              // Xr3 (b7, b6, b5, b4)
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr1,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr1,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 0);

//i = 4 - 7
    D32SLL(xr4,xr12,xr13,xr5,0);      // move (xr12, xr13) to (xr4,xr5)
    Q8ACCE_AA(xr4,xr3,xr0,xr5);    // Xr4 (b3 + dc, b2 + dc)
                                      // Xr5 (b1 + dc, b0 + dc)
    Q16SAT(xr3,xr4,xr5);              // Xr4 (b3, b2, b1, b0)
    S32STD(xr1, dst, 4);
}
#else
// assumes all AC coefs are 0
void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
    int i, j;
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
    int dc = (block[0] + 32) >> 6;
    for( j = 0; j < 8; j++ )
    {
        for( i = 0; i < 8; i++ )
            dst[i] = cm[ dst[i] + dc ];
        dst += stride;
    }
}
#endif