dct64_altivec.c

自己移植的linux下的流媒体播放器原代码,支持mms协议,支持ftp和http协议.

    
      temp1 = vec_sub(vec_perm(samplesv3, samplesv3, reverse),
                      samplesv6);
      temp2 = vec_madd(temp1,
                       vec_perm(costabv3, costabv3, reverse),
                       vczero);
      //vec_st(temp2, 80, b1);
      b1v5 = temp2;
      temp1 = vec_sub(vec_perm(samplesv2, samplesv2, reverse),
                      samplesv7);
      temp2 = vec_madd(temp1,
                       vec_perm(costabv2, costabv2, reverse),
                       vczero);
      //vec_st(temp2, 96, b1);
      b1v6 = temp2;
    
      temp1 = vec_sub(vec_perm(samplesv1, samplesv1, reverse),
                      samplesv8);
      temp2 = vec_madd(temp1,
                       vec_perm(costabv1, costabv1, reverse),
                       vczero);
      //vec_st(temp2, 112, b1);
      b1v7 = temp2;

    }

    {
      register vector float
        b2v0, b2v1, b2v2, b2v3,
        b2v4, b2v5, b2v6, b2v7;
      {
        register real *costab = mp3lib_pnts[1];
        register vector float costabv1r, costabv2r, costabv1, costabv2, costabv3;
        register vector unsigned char costab_perm = vec_lvsl(0, costab);

        costabv1 = vec_ld(0, costab);
        costabv2 = vec_ld(16, costab);
        costabv1 = vec_perm(costabv1, costabv2, costab_perm);
        costabv3  = vec_ld(32, costab);
        costabv2 = vec_perm(costabv2, costabv3 , costab_perm);
        costabv1r = vec_perm(costabv1, costabv1, reverse);
        costabv2r = vec_perm(costabv2, costabv2, reverse);
    
        temp1 = vec_add(b1v0, vec_perm(b1v3, b1v3, reverse));
        //vec_st(temp1, 0, b2);
        b2v0 = temp1;
        temp1 = vec_add(b1v1, vec_perm(b1v2, b1v2, reverse));
        //vec_st(temp1, 16, b2);
        b2v1 = temp1;
        temp2 = vec_sub(vec_perm(b1v1, b1v1, reverse), b1v2);
        temp1 = vec_madd(temp2, costabv2r, vczero);
        //vec_st(temp1, 32, b2);
        b2v2 = temp1;
        temp2 = vec_sub(vec_perm(b1v0, b1v0, reverse), b1v3);
        temp1 = vec_madd(temp2, costabv1r, vczero);
        //vec_st(temp1, 48, b2);
        b2v3 = temp1;
        temp1 = vec_add(b1v4, vec_perm(b1v7, b1v7, reverse));
        //vec_st(temp1, 64, b2);
        b2v4 = temp1;
        temp1 = vec_add(b1v5, vec_perm(b1v6, b1v6, reverse));
        //vec_st(temp1, 80, b2);
        b2v5 = temp1;
        temp2 = vec_sub(b1v6, vec_perm(b1v5, b1v5, reverse));
        temp1 = vec_madd(temp2, costabv2r, vczero);
        //vec_st(temp1, 96, b2);
        b2v6 = temp1;
        temp2 = vec_sub(b1v7, vec_perm(b1v4, b1v4, reverse));
        temp1 = vec_madd(temp2, costabv1r, vczero);
        //vec_st(temp1, 112, b2);
        b2v7 = temp1;
      }

      {
        register real *costab = mp3lib_pnts[2];

    
        vector float costabv1r, costabv1, costabv2;
        vector unsigned char costab_perm = vec_lvsl(0, costab);

        costabv1 = vec_ld(0, costab);
        costabv2 = vec_ld(16, costab);
        costabv1 = vec_perm(costabv1, costabv2, costab_perm);
        costabv1r = vec_perm(costabv1, costabv1, reverse);
    
        temp1 = vec_add(b2v0, vec_perm(b2v1, b2v1, reverse));
        vec_st(temp1, 0, b1);
        temp2 = vec_sub(vec_perm(b2v0, b2v0, reverse), b2v1);
        temp1 = vec_madd(temp2, costabv1r, vczero);
        vec_st(temp1, 16, b1);
    
        temp1 = vec_add(b2v2, vec_perm(b2v3, b2v3, reverse));
        vec_st(temp1, 32, b1);
        temp2 = vec_sub(b2v3, vec_perm(b2v2, b2v2, reverse));
        temp1 = vec_madd(temp2, costabv1r, vczero);
        vec_st(temp1, 48, b1);

        temp1 = vec_add(b2v4, vec_perm(b2v5, b2v5, reverse));
        vec_st(temp1, 64, b1);
        temp2 = vec_sub(vec_perm(b2v4, b2v4, reverse), b2v5);
        temp1 = vec_madd(temp2, costabv1r, vczero);
        vec_st(temp1, 80, b1);

        temp1 = vec_add(b2v6, vec_perm(b2v7, b2v7, reverse));
        vec_st(temp1, 96, b1);
        temp2 = vec_sub(b2v7, vec_perm(b2v6, b2v6, reverse));
        temp1 = vec_madd(temp2, costabv1r, vczero);
        vec_st(temp1, 112, b1);
    
      }
    }
  }

#endif /* ALTIVEC_USE_REFERENCE_C_CODE */

  {
    register real const cos0 = mp3lib_pnts[3][0];
    register real const cos1 = mp3lib_pnts[3][1];

    b2[0x00] = b1[0x00] + b1[0x03];
    b2[0x01] = b1[0x01] + b1[0x02];
    b2[0x02] = (b1[0x01] - b1[0x02]) * cos1;
    b2[0x03] = (b1[0x00] - b1[0x03]) * cos0;
    b2[0x04] = b1[0x04] + b1[0x07];
    b2[0x05] = b1[0x05] + b1[0x06];
    b2[0x06] = (b1[0x06] - b1[0x05]) * cos1;
    b2[0x07] = (b1[0x07] - b1[0x04]) * cos0;
    b2[0x08] = b1[0x08] + b1[0x0B];
    b2[0x09] = b1[0x09] + b1[0x0A];
    b2[0x0A] = (b1[0x09] - b1[0x0A]) * cos1;
    b2[0x0B] = (b1[0x08] - b1[0x0B]) * cos0;
    b2[0x0C] = b1[0x0C] + b1[0x0F];
    b2[0x0D] = b1[0x0D] + b1[0x0E];
    b2[0x0E] = (b1[0x0E] - b1[0x0D]) * cos1;
    b2[0x0F] = (b1[0x0F] - b1[0x0C]) * cos0;
    b2[0x10] = b1[0x10] + b1[0x13];
    b2[0x11] = b1[0x11] + b1[0x12];
    b2[0x12] = (b1[0x11] - b1[0x12]) * cos1;
    b2[0x13] = (b1[0x10] - b1[0x13]) * cos0;
    b2[0x14] = b1[0x14] + b1[0x17];
    b2[0x15] = b1[0x15] + b1[0x16];
    b2[0x16] = (b1[0x16] - b1[0x15]) * cos1;
    b2[0x17] = (b1[0x17] - b1[0x14]) * cos0;
    b2[0x18] = b1[0x18] + b1[0x1B];
    b2[0x19] = b1[0x19] + b1[0x1A];
    b2[0x1A] = (b1[0x19] - b1[0x1A]) * cos1;
    b2[0x1B] = (b1[0x18] - b1[0x1B]) * cos0;
    b2[0x1C] = b1[0x1C] + b1[0x1F];
    b2[0x1D] = b1[0x1D] + b1[0x1E];
    b2[0x1E] = (b1[0x1E] - b1[0x1D]) * cos1;
    b2[0x1F] = (b1[0x1F] - b1[0x1C]) * cos0;
  }

  {
    register real const cos0 = mp3lib_pnts[4][0];

    b1[0x00] = b2[0x00] + b2[0x01];
    b1[0x01] = (b2[0x00] - b2[0x01]) * cos0;
    b1[0x02] = b2[0x02] + b2[0x03];
    b1[0x03] = (b2[0x03] - b2[0x02]) * cos0;
    b1[0x02] += b1[0x03];

    b1[0x04] = b2[0x04] + b2[0x05];
    b1[0x05] = (b2[0x04] - b2[0x05]) * cos0;
    b1[0x06] = b2[0x06] + b2[0x07];
    b1[0x07] = (b2[0x07] - b2[0x06]) * cos0;
    b1[0x06] += b1[0x07];
    b1[0x04] += b1[0x06];
    b1[0x06] += b1[0x05];
    b1[0x05] += b1[0x07];

    b1[0x08] = b2[0x08] + b2[0x09];
    b1[0x09] = (b2[0x08] - b2[0x09]) * cos0;
    b1[0x0A] = b2[0x0A] + b2[0x0B];
    b1[0x0B] = (b2[0x0B] - b2[0x0A]) * cos0;
    b1[0x0A] += b1[0x0B];

    b1[0x0C] = b2[0x0C] + b2[0x0D];
    b1[0x0D] = (b2[0x0C] - b2[0x0D]) * cos0;
    b1[0x0E] = b2[0x0E] + b2[0x0F];
    b1[0x0F] = (b2[0x0F] - b2[0x0E]) * cos0;
    b1[0x0E] += b1[0x0F];
    b1[0x0C] += b1[0x0E];
    b1[0x0E] += b1[0x0D];
    b1[0x0D] += b1[0x0F];

    b1[0x10] = b2[0x10] + b2[0x11];
    b1[0x11] = (b2[0x10] - b2[0x11]) * cos0;
    b1[0x12] = b2[0x12] + b2[0x13];
    b1[0x13] = (b2[0x13] - b2[0x12]) * cos0;
    b1[0x12] += b1[0x13];

    b1[0x14] = b2[0x14] + b2[0x15];
    b1[0x15] = (b2[0x14] - b2[0x15]) * cos0;
    b1[0x16] = b2[0x16] + b2[0x17];
    b1[0x17] = (b2[0x17] - b2[0x16]) * cos0;
    b1[0x16] += b1[0x17];
    b1[0x14] += b1[0x16];
    b1[0x16] += b1[0x15];
    b1[0x15] += b1[0x17];

    b1[0x18] = b2[0x18] + b2[0x19];
    b1[0x19] = (b2[0x18] - b2[0x19]) * cos0;
    b1[0x1A] = b2[0x1A] + b2[0x1B];
    b1[0x1B] = (b2[0x1B] - b2[0x1A]) * cos0;
    b1[0x1A] += b1[0x1B];

    b1[0x1C] = b2[0x1C] + b2[0x1D];
    b1[0x1D] = (b2[0x1C] - b2[0x1D]) * cos0;
    b1[0x1E] = b2[0x1E] + b2[0x1F];
    b1[0x1F] = (b2[0x1F] - b2[0x1E]) * cos0;
    b1[0x1E] += b1[0x1F];
    b1[0x1C] += b1[0x1E];
    b1[0x1E] += b1[0x1D];
    b1[0x1D] += b1[0x1F];
  }

  out0[0x10*16] = b1[0x00];
  out0[0x10*12] = b1[0x04];
  out0[0x10* 8] = b1[0x02];
  out0[0x10* 4] = b1[0x06];
  out0[0x10* 0] = b1[0x01];
  out1[0x10* 0] = b1[0x01];
  out1[0x10* 4] = b1[0x05];
  out1[0x10* 8] = b1[0x03];
  out1[0x10*12] = b1[0x07];

  b1[0x08] += b1[0x0C];
  out0[0x10*14] = b1[0x08];
  b1[0x0C] += b1[0x0a];
  out0[0x10*10] = b1[0x0C];
  b1[0x0A] += b1[0x0E];
  out0[0x10* 6] = b1[0x0A];
  b1[0x0E] += b1[0x09];
  out0[0x10* 2] = b1[0x0E];
  b1[0x09] += b1[0x0D];
  out1[0x10* 2] = b1[0x09];
  b1[0x0D] += b1[0x0B];
  out1[0x10* 6] = b1[0x0D];
  b1[0x0B] += b1[0x0F];
  out1[0x10*10] = b1[0x0B];
  out1[0x10*14] = b1[0x0F];

  b1[0x18] += b1[0x1C];
  out0[0x10*15] = b1[0x10] + b1[0x18];
  out0[0x10*13] = b1[0x18] + b1[0x14];
  b1[0x1C] += b1[0x1a];
  out0[0x10*11] = b1[0x14] + b1[0x1C];
  out0[0x10* 9] = b1[0x1C] + b1[0x12];
  b1[0x1A] += b1[0x1E];
  out0[0x10* 7] = b1[0x12] + b1[0x1A];
  out0[0x10* 5] = b1[0x1A] + b1[0x16];
  b1[0x1E] += b1[0x19];
  out0[0x10* 3] = b1[0x16] + b1[0x1E];
  out0[0x10* 1] = b1[0x1E] + b1[0x11];
  b1[0x19] += b1[0x1D];
  out1[0x10* 1] = b1[0x11] + b1[0x19];
  out1[0x10* 3] = b1[0x19] + b1[0x15];
  b1[0x1D] += b1[0x1B];
  out1[0x10* 5] = b1[0x15] + b1[0x1D];
  out1[0x10* 7] = b1[0x1D] + b1[0x13];
  b1[0x1B] += b1[0x1F];
  out1[0x10* 9] = b1[0x13] + b1[0x1B];
  out1[0x10*11] = b1[0x1B] + b1[0x17];
  out1[0x10*13] = b1[0x17] + b1[0x1F];
  out1[0x10*15] = b1[0x1F];
}

#endif /* HAVE_ALTIVEC */