Exemple #1
0
static void dct64_1(real *out0,real *out1,real *b1,real *b2,real *samples)
{
 {
  register real *costab = pnts[0];

  b1[0x00] = samples[0x00] + samples[0x1F];
  b1[0x01] = samples[0x01] + samples[0x1E];
  b1[0x1F] = REAL_MUL(samples[0x00] - samples[0x1F], costab[0x0]);
  b1[0x1E] = REAL_MUL(samples[0x01] - samples[0x1E], costab[0x1]);

  b1[0x02] = samples[0x02] + samples[0x1D];
  b1[0x03] = samples[0x03] + samples[0x1C];
  b1[0x1D] = REAL_MUL(samples[0x02] - samples[0x1D], costab[0x2]);
  b1[0x1C] = REAL_MUL(samples[0x03] - samples[0x1C], costab[0x3]);

  b1[0x04] = samples[0x04] + samples[0x1B];
  b1[0x05] = samples[0x05] + samples[0x1A];
  b1[0x1B] = REAL_MUL(samples[0x04] - samples[0x1B], costab[0x4]);
  b1[0x1A] = REAL_MUL(samples[0x05] - samples[0x1A], costab[0x5]);

  b1[0x06] = samples[0x06] + samples[0x19];
  b1[0x07] = samples[0x07] + samples[0x18];
  b1[0x19] = REAL_MUL(samples[0x06] - samples[0x19], costab[0x6]);
  b1[0x18] = REAL_MUL(samples[0x07] - samples[0x18], costab[0x7]);

  b1[0x08] = samples[0x08] + samples[0x17];
  b1[0x09] = samples[0x09] + samples[0x16];
  b1[0x17] = REAL_MUL(samples[0x08] - samples[0x17], costab[0x8]);
  b1[0x16] = REAL_MUL(samples[0x09] - samples[0x16], costab[0x9]);

  b1[0x0A] = samples[0x0A] + samples[0x15];
  b1[0x0B] = samples[0x0B] + samples[0x14];
  b1[0x15] = REAL_MUL(samples[0x0A] - samples[0x15], costab[0xA]);
  b1[0x14] = REAL_MUL(samples[0x0B] - samples[0x14], costab[0xB]);

  b1[0x0C] = samples[0x0C] + samples[0x13];
  b1[0x0D] = samples[0x0D] + samples[0x12];
  b1[0x13] = REAL_MUL(samples[0x0C] - samples[0x13], costab[0xC]);
  b1[0x12] = REAL_MUL(samples[0x0D] - samples[0x12], costab[0xD]);

  b1[0x0E] = samples[0x0E] + samples[0x11];
  b1[0x0F] = samples[0x0F] + samples[0x10];
  b1[0x11] = REAL_MUL(samples[0x0E] - samples[0x11], costab[0xE]);
  b1[0x10] = REAL_MUL(samples[0x0F] - samples[0x10], costab[0xF]);

 }


 {
  register real *costab = pnts[1];

  b2[0x00] = b1[0x00] + b1[0x0F]; 
  b2[0x01] = b1[0x01] + b1[0x0E]; 
  b2[0x0F] = REAL_MUL(b1[0x00] - b1[0x0F], costab[0]);
  b2[0x0E] = REAL_MUL(b1[0x01] - b1[0x0E], costab[1]);

  b2[0x02] = b1[0x02] + b1[0x0D]; 
  b2[0x03] = b1[0x03] + b1[0x0C]; 
  b2[0x0D] = REAL_MUL(b1[0x02] - b1[0x0D], costab[2]);
  b2[0x0C] = REAL_MUL(b1[0x03] - b1[0x0C], costab[3]);

  b2[0x04] = b1[0x04] + b1[0x0B]; 
  b2[0x05] = b1[0x05] + b1[0x0A]; 
  b2[0x0B] = REAL_MUL(b1[0x04] - b1[0x0B], costab[4]);
  b2[0x0A] = REAL_MUL(b1[0x05] - b1[0x0A], costab[5]);

  b2[0x06] = b1[0x06] + b1[0x09]; 
  b2[0x07] = b1[0x07] + b1[0x08]; 
  b2[0x09] = REAL_MUL(b1[0x06] - b1[0x09], costab[6]);
  b2[0x08] = REAL_MUL(b1[0x07] - b1[0x08], costab[7]);

  /* */

  b2[0x10] = b1[0x10] + b1[0x1F];
  b2[0x11] = b1[0x11] + b1[0x1E];
  b2[0x1F] = REAL_MUL(b1[0x1F] - b1[0x10], costab[0]);
  b2[0x1E] = REAL_MUL(b1[0x1E] - b1[0x11], costab[1]);

  b2[0x12] = b1[0x12] + b1[0x1D];
  b2[0x13] = b1[0x13] + b1[0x1C];
  b2[0x1D] = REAL_MUL(b1[0x1D] - b1[0x12], costab[2]);
  b2[0x1C] = REAL_MUL(b1[0x1C] - b1[0x13], costab[3]);

  b2[0x14] = b1[0x14] + b1[0x1B];
  b2[0x15] = b1[0x15] + b1[0x1A];
  b2[0x1B] = REAL_MUL(b1[0x1B] - b1[0x14], costab[4]);
  b2[0x1A] = REAL_MUL(b1[0x1A] - b1[0x15], costab[5]);

  b2[0x16] = b1[0x16] + b1[0x19];
  b2[0x17] = b1[0x17] + b1[0x18];
  b2[0x19] = REAL_MUL(b1[0x19] - b1[0x16], costab[6]);
  b2[0x18] = REAL_MUL(b1[0x18] - b1[0x17], costab[7]);
 }

 {
  register real *costab = pnts[2];

  b1[0x00] = b2[0x00] + b2[0x07];
  b1[0x07] = REAL_MUL(b2[0x00] - b2[0x07], costab[0]);
  b1[0x01] = b2[0x01] + b2[0x06];
  b1[0x06] = REAL_MUL(b2[0x01] - b2[0x06], costab[1]);
  b1[0x02] = b2[0x02] + b2[0x05];
  b1[0x05] = REAL_MUL(b2[0x02] - b2[0x05], costab[2]);
  b1[0x03] = b2[0x03] + b2[0x04];
  b1[0x04] = REAL_MUL(b2[0x03] - b2[0x04], costab[3]);

  b1[0x08] = b2[0x08] + b2[0x0F];
  b1[0x0F] = REAL_MUL(b2[0x0F] - b2[0x08], costab[0]);
  b1[0x09] = b2[0x09] + b2[0x0E];
  b1[0x0E] = REAL_MUL(b2[0x0E] - b2[0x09], costab[1]);
  b1[0x0A] = b2[0x0A] + b2[0x0D];
  b1[0x0D] = REAL_MUL(b2[0x0D] - b2[0x0A], costab[2]);
  b1[0x0B] = b2[0x0B] + b2[0x0C];
  b1[0x0C] = REAL_MUL(b2[0x0C] - b2[0x0B], costab[3]);

  b1[0x10] = b2[0x10] + b2[0x17];
  b1[0x17] = REAL_MUL(b2[0x10] - b2[0x17], costab[0]);
  b1[0x11] = b2[0x11] + b2[0x16];
  b1[0x16] = REAL_MUL(b2[0x11] - b2[0x16], costab[1]);
  b1[0x12] = b2[0x12] + b2[0x15];
  b1[0x15] = REAL_MUL(b2[0x12] - b2[0x15], costab[2]);
  b1[0x13] = b2[0x13] + b2[0x14];
  b1[0x14] = REAL_MUL(b2[0x13] - b2[0x14], costab[3]);

  b1[0x18] = b2[0x18] + b2[0x1F];
  b1[0x1F] = REAL_MUL(b2[0x1F] - b2[0x18], costab[0]);
  b1[0x19] = b2[0x19] + b2[0x1E];
  b1[0x1E] = REAL_MUL(b2[0x1E] - b2[0x19], costab[1]);
  b1[0x1A] = b2[0x1A] + b2[0x1D];
  b1[0x1D] = REAL_MUL(b2[0x1D] - b2[0x1A], costab[2]);
  b1[0x1B] = b2[0x1B] + b2[0x1C];
  b1[0x1C] = REAL_MUL(b2[0x1C] - b2[0x1B], costab[3]);
 }

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

  b2[0x00] = b1[0x00] + b1[0x03];
  b2[0x03] = REAL_MUL(b1[0x00] - b1[0x03], cos0);
  b2[0x01] = b1[0x01] + b1[0x02];
  b2[0x02] = REAL_MUL(b1[0x01] - b1[0x02], cos1);

  b2[0x04] = b1[0x04] + b1[0x07];
  b2[0x07] = REAL_MUL(b1[0x07] - b1[0x04], cos0);
  b2[0x05] = b1[0x05] + b1[0x06];
  b2[0x06] = REAL_MUL(b1[0x06] - b1[0x05], cos1);

  b2[0x08] = b1[0x08] + b1[0x0B];
  b2[0x0B] = REAL_MUL(b1[0x08] - b1[0x0B], cos0);
  b2[0x09] = b1[0x09] + b1[0x0A];
  b2[0x0A] = REAL_MUL(b1[0x09] - b1[0x0A], cos1);
  
  b2[0x0C] = b1[0x0C] + b1[0x0F];
  b2[0x0F] = REAL_MUL(b1[0x0F] - b1[0x0C], cos0);
  b2[0x0D] = b1[0x0D] + b1[0x0E];
  b2[0x0E] = REAL_MUL(b1[0x0E] - b1[0x0D], cos1);

  b2[0x10] = b1[0x10] + b1[0x13];
  b2[0x13] = REAL_MUL(b1[0x10] - b1[0x13], cos0);
  b2[0x11] = b1[0x11] + b1[0x12];
  b2[0x12] = REAL_MUL(b1[0x11] - b1[0x12], cos1);

  b2[0x14] = b1[0x14] + b1[0x17];
  b2[0x17] = REAL_MUL(b1[0x17] - b1[0x14], cos0);
  b2[0x15] = b1[0x15] + b1[0x16];
  b2[0x16] = REAL_MUL(b1[0x16] - b1[0x15], cos1);

  b2[0x18] = b1[0x18] + b1[0x1B];
  b2[0x1B] = REAL_MUL(b1[0x18] - b1[0x1B], cos0);
  b2[0x19] = b1[0x19] + b1[0x1A];
  b2[0x1A] = REAL_MUL(b1[0x19] - b1[0x1A], cos1);

  b2[0x1C] = b1[0x1C] + b1[0x1F];
  b2[0x1F] = REAL_MUL(b1[0x1F] - b1[0x1C], cos0);
  b2[0x1D] = b1[0x1D] + b1[0x1E];
  b2[0x1E] = REAL_MUL(b1[0x1E] - b1[0x1D], cos1);
 }

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

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

  b1[0x04] = b2[0x04] + b2[0x05];
  b1[0x05] = REAL_MUL(b2[0x04] - b2[0x05], cos0);
  b1[0x06] = b2[0x06] + b2[0x07];
  b1[0x07] = REAL_MUL(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] = REAL_MUL(b2[0x08] - b2[0x09], cos0);
  b1[0x0A] = b2[0x0A] + b2[0x0B];
  b1[0x0B] = REAL_MUL(b2[0x0B] - b2[0x0A], cos0);
  b1[0x0A] += b1[0x0B];

  b1[0x0C] = b2[0x0C] + b2[0x0D];
  b1[0x0D] = REAL_MUL(b2[0x0C] - b2[0x0D], cos0);
  b1[0x0E] = b2[0x0E] + b2[0x0F];
  b1[0x0F] = REAL_MUL(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] = REAL_MUL(b2[0x10] - b2[0x11], cos0);
  b1[0x12] = b2[0x12] + b2[0x13];
  b1[0x13] = REAL_MUL(b2[0x13] - b2[0x12], cos0);
  b1[0x12] += b1[0x13];

  b1[0x14] = b2[0x14] + b2[0x15];
  b1[0x15] = REAL_MUL(b2[0x14] - b2[0x15], cos0);
  b1[0x16] = b2[0x16] + b2[0x17];
  b1[0x17] = REAL_MUL(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] = REAL_MUL(b2[0x18] - b2[0x19], cos0);
  b1[0x1A] = b2[0x1A] + b2[0x1B];
  b1[0x1B] = REAL_MUL(b2[0x1B] - b2[0x1A], cos0);
  b1[0x1A] += b1[0x1B];

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

 out0[0x10*16] = REAL_SCALE_DCT64(b1[0x00]);
 out0[0x10*12] = REAL_SCALE_DCT64(b1[0x04]);
 out0[0x10* 8] = REAL_SCALE_DCT64(b1[0x02]);
 out0[0x10* 4] = REAL_SCALE_DCT64(b1[0x06]);
 out0[0x10* 0] = REAL_SCALE_DCT64(b1[0x01]);
 out1[0x10* 0] = REAL_SCALE_DCT64(b1[0x01]);
 out1[0x10* 4] = REAL_SCALE_DCT64(b1[0x05]);
 out1[0x10* 8] = REAL_SCALE_DCT64(b1[0x03]);
 out1[0x10*12] = REAL_SCALE_DCT64(b1[0x07]);

#if 1
 out0[0x10*14] = REAL_SCALE_DCT64(b1[0x08] + b1[0x0C]);
 out0[0x10*10] = REAL_SCALE_DCT64(b1[0x0C] + b1[0x0a]);
 out0[0x10* 6] = REAL_SCALE_DCT64(b1[0x0A] + b1[0x0E]);
 out0[0x10* 2] = REAL_SCALE_DCT64(b1[0x0E] + b1[0x09]);
 out1[0x10* 2] = REAL_SCALE_DCT64(b1[0x09] + b1[0x0D]);
 out1[0x10* 6] = REAL_SCALE_DCT64(b1[0x0D] + b1[0x0B]);
 out1[0x10*10] = REAL_SCALE_DCT64(b1[0x0B] + b1[0x0F]);
 out1[0x10*14] = REAL_SCALE_DCT64(b1[0x0F]);
#else
 b1[0x08] += b1[0x0C];
 out0[0x10*14] = REAL_SCALE_DCT64(b1[0x08]);
 b1[0x0C] += b1[0x0a];
 out0[0x10*10] = REAL_SCALE_DCT64(b1[0x0C]);
 b1[0x0A] += b1[0x0E];
 out0[0x10* 6] = REAL_SCALE_DCT64(b1[0x0A]);
 b1[0x0E] += b1[0x09];
 out0[0x10* 2] = REAL_SCALE_DCT64(b1[0x0E]);
 b1[0x09] += b1[0x0D];
 out1[0x10* 2] = REAL_SCALE_DCT64(b1[0x09]);
 b1[0x0D] += b1[0x0B];
 out1[0x10* 6] = REAL_SCALE_DCT64(b1[0x0D]);
 b1[0x0B] += b1[0x0F];
 out1[0x10*10] = REAL_SCALE_DCT64(b1[0x0B]);
 out1[0x10*14] = REAL_SCALE_DCT64(b1[0x0F]);
#endif

 { 
 real tmp;
 tmp = b1[0x18] + b1[0x1C];
 out0[0x10*15] = REAL_SCALE_DCT64(tmp + b1[0x10]);
 out0[0x10*13] = REAL_SCALE_DCT64(tmp + b1[0x14]);
 tmp = b1[0x1C] + b1[0x1A];
 out0[0x10*11] = REAL_SCALE_DCT64(tmp + b1[0x14]);
 out0[0x10* 9] = REAL_SCALE_DCT64(tmp + b1[0x12]);
 tmp = b1[0x1A] + b1[0x1E];
 out0[0x10* 7] = REAL_SCALE_DCT64(tmp + b1[0x12]);
 out0[0x10* 5] = REAL_SCALE_DCT64(tmp + b1[0x16]);
 tmp = b1[0x1E] + b1[0x19];
 out0[0x10* 3] = REAL_SCALE_DCT64(tmp + b1[0x16]);
 out0[0x10* 1] = REAL_SCALE_DCT64(tmp + b1[0x11]);
 tmp = b1[0x19] + b1[0x1D];
 out1[0x10* 1] = REAL_SCALE_DCT64(tmp + b1[0x11]);
 out1[0x10* 3] = REAL_SCALE_DCT64(tmp + b1[0x15]); 
 tmp = b1[0x1D] + b1[0x1B];
 out1[0x10* 5] = REAL_SCALE_DCT64(tmp + b1[0x15]);
 out1[0x10* 7] = REAL_SCALE_DCT64(tmp + b1[0x13]);
 tmp = b1[0x1B] + b1[0x1F];
 out1[0x10* 9] = REAL_SCALE_DCT64(tmp + b1[0x13]);
 out1[0x10*11] = REAL_SCALE_DCT64(tmp + b1[0x17]);
 out1[0x10*13] = REAL_SCALE_DCT64(b1[0x17] + b1[0x1F]);
 out1[0x10*15] = REAL_SCALE_DCT64(b1[0x1F]);
 }
}
void dct64(real *out0,real *out1,real *samples)
{
  real bufs[64];

 {
  register int i,j;
  register real *b1,*b2,*bs,*costab;

  b1 = samples;
  bs = bufs;
  costab = pnts[0]+16;
  b2 = b1 + 32; 
/*
//C Code to run IDCT Core Hardware
	int count;
	unsigned int stat,idctfin;

	count=32;
	stat=idct_regs->flag_read_stat;
	stat=stat%4;
	if (stat==3 || stat==2 || stat==1){
			while (count){
				b2--;
				idct_regs->flag_send_data=*b2;
				count--;
				}
  			}	
	idct_regs->idct_write_reset;
	idctfin=1;	
	while (idctfin){
		idctfin=idct_regs->flag_idct_finished;
		idctfin=idctfin%4;
		};
	count=32;
	while(count){
		*bs=idct_regs->flag_read_data;
		count--;
		bs++;}
}
*/
//Removed C code which is implemented in IDCT Core Hardware

  for(i=15;i>=0;i--)
    *bs++ = (*b1++ + *--b2); 
  for(i=15;i>=0;i--)
    *bs++ = REAL_MUL((*--b2 - *b1++), *--costab);

  b1 = bufs;
  costab = pnts[1]+8;
  b2 = b1 + 16;

  {
    for(i=7;i>=0;i--)
      *bs++ = (*b1++ + *--b2); 
    for(i=7;i>=0;i--)
      *bs++ = REAL_MUL((*--b2 - *b1++), *--costab);
    b2 += 32;
    costab += 8;
    for(i=7;i>=0;i--)
      *bs++ = (*b1++ + *--b2); 
    for(i=7;i>=0;i--)
      *bs++ = REAL_MUL((*b1++ - *--b2), *--costab);
    b2 += 32;
  }

  bs = bufs;
  costab = pnts[2];
  b2 = b1 + 8;

  for(j=2;j;j--)
  {
    for(i=3;i>=0;i--)
      *bs++ = (*b1++ + *--b2); 
    for(i=3;i>=0;i--)
      *bs++ = REAL_MUL((*--b2 - *b1++), costab[i]);
    b2 += 16;
    for(i=3;i>=0;i--)
      *bs++ = (*b1++ + *--b2); 
    for(i=3;i>=0;i--)
      *bs++ = REAL_MUL((*b1++ - *--b2), costab[i]);
    b2 += 16;
  }

  b1 = bufs;
  costab = pnts[3];
  b2 = b1 + 4;

  for(j=4;j;j--)
  {
    *bs++ = (*b1++ + *--b2); 
    *bs++ = (*b1++ + *--b2);
    *bs++ = REAL_MUL((*--b2 - *b1++), costab[1]);
    *bs++ = REAL_MUL((*--b2 - *b1++), costab[0]);
    b2 += 8;
    *bs++ = (*b1++ + *--b2); 
    *bs++ = (*b1++ + *--b2);
    *bs++ = REAL_MUL((*b1++ - *--b2), costab[1]);
    *bs++ = REAL_MUL((*b1++ - *--b2), costab[0]);
    b2 += 8;
  }
  bs = bufs;
  costab = pnts[4];

  for(j=8;j;j--)
  {
    real v0,v1;
    v0=*b1++; v1 = *b1++;
    *bs++ = (v0 + v1);
    *bs++ = REAL_MUL((v0 - v1), (*costab));
    v0=*b1++; v1 = *b1++;
    *bs++ = (v0 + v1);
    *bs++ = REAL_MUL((v1 - v0), (*costab));
  }

 }


 {
  register real *b1;
  register int i;

  for(b1=bufs,i=8;i;i--,b1+=4)
    b1[2] += b1[3];

  for(b1=bufs,i=4;i;i--,b1+=8)
  {
    b1[4] += b1[6];
    b1[6] += b1[5];
    b1[5] += b1[7];
  }

  for(b1=bufs,i=2;i;i--,b1+=16)
  {
    b1[8]  += b1[12];
    b1[12] += b1[10];
    b1[10] += b1[14];
    b1[14] += b1[9];
    b1[9]  += b1[13];
    b1[13] += b1[11];
    b1[11] += b1[15];
  }
 }


  out0[0x10*16] = REAL_SCALE_DCT64(bufs[0]);
  out0[0x10*15] = REAL_SCALE_DCT64(bufs[16+0]  + bufs[16+8]);
  out0[0x10*14] = REAL_SCALE_DCT64(bufs[8]);
  out0[0x10*13] = REAL_SCALE_DCT64(bufs[16+8]  + bufs[16+4]);
  out0[0x10*12] = REAL_SCALE_DCT64(bufs[4]);
  out0[0x10*11] = REAL_SCALE_DCT64(bufs[16+4]  + bufs[16+12]);
  out0[0x10*10] = REAL_SCALE_DCT64(bufs[12]);
  out0[0x10* 9] = REAL_SCALE_DCT64(bufs[16+12] + bufs[16+2]);
  out0[0x10* 8] = REAL_SCALE_DCT64(bufs[2]);
  out0[0x10* 7] = REAL_SCALE_DCT64(bufs[16+2]  + bufs[16+10]);
  out0[0x10* 6] = REAL_SCALE_DCT64(bufs[10]);
  out0[0x10* 5] = REAL_SCALE_DCT64(bufs[16+10] + bufs[16+6]);
  out0[0x10* 4] = REAL_SCALE_DCT64(bufs[6]);
  out0[0x10* 3] = REAL_SCALE_DCT64(bufs[16+6]  + bufs[16+14]);
  out0[0x10* 2] = REAL_SCALE_DCT64(bufs[14]);
  out0[0x10* 1] = REAL_SCALE_DCT64(bufs[16+14] + bufs[16+1]);
  out0[0x10* 0] = REAL_SCALE_DCT64(bufs[1]);

  out1[0x10* 0] = REAL_SCALE_DCT64(bufs[1]);
  out1[0x10* 1] = REAL_SCALE_DCT64(bufs[16+1]  + bufs[16+9]);
  out1[0x10* 2] = REAL_SCALE_DCT64(bufs[9]);
  out1[0x10* 3] = REAL_SCALE_DCT64(bufs[16+9]  + bufs[16+5]);
  out1[0x10* 4] = REAL_SCALE_DCT64(bufs[5]);
  out1[0x10* 5] = REAL_SCALE_DCT64(bufs[16+5]  + bufs[16+13]);
  out1[0x10* 6] = REAL_SCALE_DCT64(bufs[13]);
  out1[0x10* 7] = REAL_SCALE_DCT64(bufs[16+13] + bufs[16+3]);
  out1[0x10* 8] = REAL_SCALE_DCT64(bufs[3]);
  out1[0x10* 9] = REAL_SCALE_DCT64(bufs[16+3]  + bufs[16+11]);
  out1[0x10*10] = REAL_SCALE_DCT64(bufs[11]);
  out1[0x10*11] = REAL_SCALE_DCT64(bufs[16+11] + bufs[16+7]);
  out1[0x10*12] = REAL_SCALE_DCT64(bufs[7]);
  out1[0x10*13] = REAL_SCALE_DCT64(bufs[16+7]  + bufs[16+15]);
  out1[0x10*14] = REAL_SCALE_DCT64(bufs[15]);
  out1[0x10*15] = REAL_SCALE_DCT64(bufs[16+15]);

}