mp3enc_layer3_int.cpp

这是在PCA下的基于IPP库示例代码例子,在网上下了IPP的库之后,设置相关参数就可以编译该代码.

/*//////////////////////////////////////////////////////////////////////////////
//
//                  INTEL CORPORATION PROPRIETARY INFORMATION
//     This software is supplied under the terms of a license agreement or
//     nondisclosure agreement with Intel Corporation and may not be copied
//     or disclosed except in accordance with the terms of that agreement.
//          Copyright(c) 2002-2005 Intel Corporation. All Rights Reserved.
//
*/

#include "umc_mp3_enc_int.h"

namespace UMC {

/* Coefficient Wi * (2/3)^2 * sqrt(2) in Q31 */
static const Ipp32s pIMWin[4][36] =
{
    {
        58876489, 176181383, 292145429, 405886073, 516537678,
        623258120, 725235191, 821692784, 911896799, 995160729,
        1070850885, 1138391219, 1197267708, 1247032268, 1287306158,
        1317782872, 1338230462, 1348493310, 1348493310, 1338230462,
        1317782872, 1287306158, 1247032268, 1197267708, 1138391219,
        1070850885, 995160729, 911896799, 821692784, 725235191,
        623258120, 516537678, 405886073, 292145429, 176181383, 58876489
    },

    {
        58876489, 176181383, 292145429, 405886073, 516537678, 623258120,
        725235191, 821692784, 911896799, 995160729, 1070850885, 1138391219,
        1197267708, 1247032268, 1287306158, 1317782872, 1338230462, 1348493310,
        1349778000, 1349778000, 1349778000, 1349778000, 1349778000, 1349778000,
        1338230462, 1247032268, 1070850885, 821692784, 516537678, 176181383,
        0,0,0,0,0,0
    },

    {
        176181383, 516537678, 821692784, 1070850885, 1247032268, 1338230462,
        1338230462, 1247032268, 1070850885, 821692784, 516537678, 176181383,
        0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
    },

    {
        0,0,0,0,0,0,
        176181383,516537678,821692784,1070850885,1247032268,1338230462,
        1349778000,1349778000,1349778000,1349778000,1349778000,1349778000,
        1348493310,1338230462,1317782872,1287306158,1247032268,1197267708,
        1138391219,1070850885,995160729,911896799,821692784,725235191,623258120,
        516537678,405886073,292145429,176181383,58876489
    }
};


/*
//  cos(17*pi/72)  cos(15*pi/72)   cos(13*pi/72)   cos(11*pi/72)
//  cos( 9*pi/72)  cos( 7*pi/72)   cos( 5*pi/72)   cos( 3*pi/72)
//  cos(   pi/72)
//  in Q31
*/

static const Ipp32s _d72cos[9] =
{
    1583291025, 1703713325, 1811169339, 1904841260,
    1984016189, 2048091557, 2096579711, 2129111628,
    2145439719
};

/*
//  sin(17*pi/72)  sin(15*pi/72)   sin(13*pi/72)   sin(11*pi/72)
//  sin( 9*pi/72)  sin( 7*pi/72)   sin( 5*pi/72)   sin( 3*pi/72)
//  sin(   pi/72)
//  in Q31
*/

static const Ipp32s _d72sin[9] =
{
    1450818924, 1307305214, 1153842123, 991597596,
    821806413, 645760787, 464800532, 280302863,
    93671921
};


/* coefficient in Q31 */
static const Ipp32s _dt[8] =
{
    1859775393, 1073741824, -2017974537, 372906622,
    1645067915, 734482665, 2114858546, 1380375881
};


/*
//  cos(pi/24) cos(3*pi/24) cos(5*pi/24)
//  sin(pi/24) sin(3*pi/24) sin(5*pi/24)
//  in Q31
*/

static const Ipp32s _dcoef12[6] =
{
    2129111628, 1984016189, 1703713325, 280302863, 821806413, 1307305214
};


/*/////////////////////////////////////////////////////////////////////////////
// Name:             ownsMDCT_36_32s
// Description:      Performs 36 points windowing and Forward MDCT for long block.
// Input Arguments : pSrc           - pointer to the input buffer which is the
//                                    output of windowing.
//                   blockType      - indication of the window type:
//                                  - 0: normal window; 1: start block;
//                                  - 2: short block;   3: stop block.
// Output Arguments: pDst           - pointer to the output buffer of Forward MDCT
//                                    for long block.
******************************************************************************/

static void ownsMDCT_36_32s(
  Ipp32s* pSrc,
  Ipp32s* pDst,
  int     blockType)
{
  int n;
  Ipp32s y[18], x[18], xa[9], a[27], m[11];

  {
    Ipp32s data[36], data2[36];

    /* phase 1 */
    for( n = 0; n < 36; n += 4 ) {
        data[n+0] = _IPPS_MUL32_MP3_32S(pSrc[n+0], pIMWin[blockType][n+0]);
        data[n+1] = _IPPS_MUL32_MP3_32S(pSrc[n+1], pIMWin[blockType][n+1]);
        data[n+2] = _IPPS_MUL32_MP3_32S(pSrc[n+2], pIMWin[blockType][n+2]);
        data[n+3] = _IPPS_MUL32_MP3_32S(pSrc[n+3], pIMWin[blockType][n+3]);
    }

    /* phase 2 */
    for( n = 0; n < 18; n += 2 )
    {
        data2[n]    = data[n]   + data[35-n];
        data2[35-n] = data[n]   - data[35-n];
        data2[n+1]  = data[n+1] + data[34-n];
        data2[34-n] = data[n+1] - data[34-n];
    }

    /* phase 3 */
    for( n = 0; n < 9; n++ )
    {
        y[n]   = data2[27+n] - data2[9+n];
        y[9+n] = data2[n]    - data2[18+n];
    }

    /* phase 4 */
    for( n = 0; n < 9; n++ )
    {
        x[n]   = (Ipp32s)(((Ipp64s)y[n]    * _d72cos[n] -
                           (Ipp64s)y[17-n] * _d72sin[n]) >> 31);
        x[9+n] = (Ipp32s)(((Ipp64s)y[9+n]  * _d72cos[8-n] +
                           (Ipp64s)y[8-n]  * _d72sin[8-n])>>31);
    }
  }

  /* phase 5: 9 points DCT-II & DST-II */
  /* Reverse Order of Input of DCT-II */
  for( n = 0; n < 9; n++ )
  {
    xa[n] = x[8-n];
  }

  /* 9 points DCT-II */
  a[1] = xa[3] + xa[5];
  a[2] = xa[3] - xa[5];
  a[3] = xa[6] + xa[2];
  a[4] = xa[6] - xa[2];
  a[5] = xa[1] + xa[7];
  a[6] = xa[1] - xa[7];
  a[7] = xa[8] + xa[0];
  a[8] = xa[8] - xa[0];

  a[9] = xa[4] + a[5];
  a[10] = a[1] + a[3];
  a[11] = a[10] + a[7];
  a[12] = a[3] - a[7];
  a[13] = a[1] - a[7];
  a[14] = a[1] - a[3];
  a[15] = a[2] - a[4];
  a[16] = a[15] + a[8];
  a[17] = a[4] + a[8];
  a[18] = a[2] - a[8];
  a[19] = a[2] + a[4];

  m[1]  = (Ipp32s)(((Ipp64s) - _dt[0] * a[6])>>31);
  m[2]  = (Ipp32s)(((Ipp64s)   _dt[1] * a[5])>>31);
  m[3]  = (Ipp32s)(((Ipp64s)   _dt[1] * a[11])>>31);
  m[4]  = (Ipp32s)(((Ipp64s) - _dt[2] * a[12])>>31);
  m[5]  = (Ipp32s)(((Ipp64s) - _dt[3] * a[13])>>31);
  m[6]  = (Ipp32s)(((Ipp64s) - _dt[4] * a[14])>>31);
  m[7]  = (Ipp32s)(((Ipp64s) - _dt[0] * a[16])>>31);
  m[8]  = (Ipp32s)(((Ipp64s) - _dt[5] * a[17])>>31);
  m[9]  = (Ipp32s)(((Ipp64s) - _dt[6] * a[18])>>31);
  m[10] = (Ipp32s)(((Ipp64s) - _dt[7] * a[19])>>31);

  a[20] = xa[4] - m[2];
  a[21] = a[20] + m[4];
  a[22] = a[20] - m[4];
  a[23] = a[20] + m[5];
  a[24] = m[1] + m[8];
  a[25] = m[1] - m[8];
  a[26] = m[1] + m[9];

  y[0] = a[9] + a[11];
  y[1] = m[10] - a[26];
  y[2] = m[6] - a[21];
  y[3] = m[7];
  y[4] = a[22] - m[5];
  y[5] = a[25] - m[9];
  y[6] = m[3] - a[9];
  y[7] = a[24] + m[10];
  y[8] = a[23] + m[6];

  /* Keep Order of Input of DST-II */
  for( n = 0; n < 9; n++ )
  {
    xa[n] = x[9+n];
  }

  xa[1] = -xa[1];
  xa[3] = -xa[3];
  xa[5] = -xa[5];
  xa[7] = -xa[7];

  /* 9 points DCT-II */
  a[1] = xa[3] + xa[5];
  a[2] = xa[3] - xa[5];
  a[3] = xa[6] + xa[2];
  a[4] = xa[6] - xa[2];
  a[5] = xa[1] + xa[7];
  a[6] = xa[1] - xa[7];
  a[7] = xa[8] + xa[0];
  a[8] = xa[8] - xa[0];

  a[9] = xa[4] + a[5];
  a[10] = a[1] + a[3];
  a[11] = a[10] + a[7];
  a[12] = a[3] - a[7];
  a[13] = a[1] - a[7];
  a[14] = a[1] - a[3];
  a[15] = a[2] - a[4];
  a[16] = a[15] + a[8];
  a[17] = a[4] + a[8];
  a[18] = a[2] - a[8];
  a[19] = a[2] + a[4];

  m[1] = (Ipp32s)(((Ipp64s)-_dt[0] * a[6])>>31);
  m[2] = (Ipp32s)(((Ipp64s)_dt[1] * a[5])>>31);
  m[3] = (Ipp32s)(((Ipp64s)_dt[1] * a[11])>>31);
  m[4] = (Ipp32s)(((Ipp64s)-_dt[2] * a[12])>>31);
  m[5] = (Ipp32s)(((Ipp64s)-_dt[3] * a[13])>>31);
  m[6] = (Ipp32s)(((Ipp64s)-_dt[4] * a[14])>>31);
  m[7] = (Ipp32s)(((Ipp64s)-_dt[0] * a[16])>>31);
  m[8] = (Ipp32s)(((Ipp64s)-_dt[5] * a[17])>>31);
  m[9] = (Ipp32s)(((Ipp64s)-_dt[6] * a[18])>>31);
  m[10] = (Ipp32s)(((Ipp64s)-_dt[7] * a[19])>>31);

  a[20] = xa[4] - m[2];
  a[21] = a[20] + m[4];
  a[22] = a[20] - m[4];
  a[23] = a[20] + m[5];
  a[24] = m[1] + m[8];
  a[25] = m[1] - m[8];
  a[26] = m[1] + m[9];

  /* Reverse output order of DST-II */
  y[9] = a[9] + a[11];
  y[10] = m[10] - a[26];
  y[11] = m[6] - a[21];
  y[12] = m[7];
  y[13] = a[22] - m[5];
  y[14] = a[25] - m[9];
  y[15] = m[3] - a[9];
  y[16] = a[24] + m[10];
  y[17] = a[23] + m[6];

  /* phase 6: output */
  pDst[0] = y[0];
  pDst[1] = -y[1] - y[17];
  pDst[2] = -y[1] + y[17];
  pDst[3] = y[2] + y[16];
  pDst[4] = y[2] - y[16];
  pDst[5] = -y[3] - y[15];
  pDst[6] = -y[3] + y[15];
  pDst[7] = y[4] + y[14];
  pDst[8] = y[4] - y[14];
  pDst[9] = -y[5] - y[13];
  pDst[10] = -y[5] + y[13];
  pDst[11] = y[6] + y[12];
  pDst[12] = y[6] - y[12];
  pDst[13] = -y[7] - y[11];
  pDst[14] = -y[7] + y[11];
  pDst[15] = y[8] + y[10];
  pDst[16] = y[8] - y[10];