Пример #1
0
jpeg_fdct_islow (DCTELEM * data)
{
  INT32 tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
  INT32 tmp10, tmp11, tmp12, tmp13;
  INT32 z1, z2, z3, z4, z5;
  DCTELEM *dataptr;
  int ctr;
  SHIFT_TEMPS

  /* Pass 1: process rows. */
  /* Note results are scaled up by sqrt(8) compared to a true DCT; */
  /* furthermore, we scale the results by 2**PASS1_BITS. */

  dataptr = data;
  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
    tmp0 = dataptr[0] + dataptr[7];
    tmp7 = dataptr[0] - dataptr[7];
    tmp1 = dataptr[1] + dataptr[6];
    tmp6 = dataptr[1] - dataptr[6];
    tmp2 = dataptr[2] + dataptr[5];
    tmp5 = dataptr[2] - dataptr[5];
    tmp3 = dataptr[3] + dataptr[4];
    tmp4 = dataptr[3] - dataptr[4];

    /* Even part per LL&M figure 1 --- note that published figure is faulty;
     * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
     */

    tmp10 = tmp0 + tmp3;
    tmp13 = tmp0 - tmp3;
    tmp11 = tmp1 + tmp2;
    tmp12 = tmp1 - tmp2;

    dataptr[0] = (DCTELEM) LEFT_SHIFT(tmp10 + tmp11, PASS1_BITS);
    dataptr[4] = (DCTELEM) LEFT_SHIFT(tmp10 - tmp11, PASS1_BITS);

    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
    dataptr[2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
                                   CONST_BITS-PASS1_BITS);
    dataptr[6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
                                   CONST_BITS-PASS1_BITS);

    /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
     * cK represents cos(K*pi/16).
     * i0..i3 in the paper are tmp4..tmp7 here.
     */

    z1 = tmp4 + tmp7;
    z2 = tmp5 + tmp6;
    z3 = tmp4 + tmp6;
    z4 = tmp5 + tmp7;
    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */

    tmp4 = MULTIPLY(tmp4, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
    tmp5 = MULTIPLY(tmp5, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
    tmp6 = MULTIPLY(tmp6, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
    tmp7 = MULTIPLY(tmp7, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */

    z3 += z5;
    z4 += z5;

    dataptr[7] = (DCTELEM) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
    dataptr[5] = (DCTELEM) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
    dataptr[3] = (DCTELEM) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
    dataptr[1] = (DCTELEM) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);

    dataptr += DCTSIZE;         /* advance pointer to next row */
  }

  /* Pass 2: process columns.
   * We remove the PASS1_BITS scaling, but leave the results scaled up
   * by an overall factor of 8.
   */

  dataptr = data;
  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
    tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
    tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
    tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
    tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
    tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
    tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
    tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
    tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];

    /* Even part per LL&M figure 1 --- note that published figure is faulty;
     * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
     */

    tmp10 = tmp0 + tmp3;
    tmp13 = tmp0 - tmp3;
    tmp11 = tmp1 + tmp2;
    tmp12 = tmp1 - tmp2;

    dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp11, PASS1_BITS);
    dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp10 - tmp11, PASS1_BITS);

    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
    dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
                                           CONST_BITS+PASS1_BITS);
    dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
                                           CONST_BITS+PASS1_BITS);

    /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
     * cK represents cos(K*pi/16).
     * i0..i3 in the paper are tmp4..tmp7 here.
     */

    z1 = tmp4 + tmp7;
    z2 = tmp5 + tmp6;
    z3 = tmp4 + tmp6;
    z4 = tmp5 + tmp7;
    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */

    tmp4 = MULTIPLY(tmp4, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
    tmp5 = MULTIPLY(tmp5, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
    tmp6 = MULTIPLY(tmp6, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
    tmp7 = MULTIPLY(tmp7, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */

    z3 += z5;
    z4 += z5;

    dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp4 + z1 + z3,
                                           CONST_BITS+PASS1_BITS);
    dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp5 + z2 + z4,
                                           CONST_BITS+PASS1_BITS);
    dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp6 + z2 + z3,
                                           CONST_BITS+PASS1_BITS);
    dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp7 + z1 + z4,
                                           CONST_BITS+PASS1_BITS);

    dataptr++;                  /* advance pointer to next column */
  }
}
Пример #2
0
jpeg_idct_4x4 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
	       JCOEFPTR coef_block,
	       JSAMPARRAY output_buf, JDIMENSION output_col)
{
  INT32 tmp0, tmp2, tmp10, tmp12;
  INT32 z1, z2, z3, z4;
  JCOEFPTR inptr;
  ISLOW_MULT_TYPE * quantptr;
  int * wsptr;
  JSAMPROW outptr;
  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
  int ctr;
  int workspace[DCTSIZE*4];	/* buffers data between passes */
  SHIFT_TEMPS

  /* Pass 1: process columns from input, store into work array. */

  inptr = coef_block;
  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
  wsptr = workspace;
  for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) {
    /* Don't bother to process column 4, because second pass won't use it */
    if (ctr == DCTSIZE-4)
      continue;
    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
	inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*5] == 0 &&
	inptr[DCTSIZE*6] == 0 && inptr[DCTSIZE*7] == 0) {
      /* AC terms all zero; we need not examine term 4 for 4x4 output */
      int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
      
      wsptr[DCTSIZE*0] = dcval;
      wsptr[DCTSIZE*1] = dcval;
      wsptr[DCTSIZE*2] = dcval;
      wsptr[DCTSIZE*3] = dcval;
      
      continue;
    }
    
    /* Even part */
    
    tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
    tmp0 <<= (CONST_BITS+1);
    
    z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
    z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);

    tmp2 = MULTIPLY(z2, FIX_1_847759065) + MULTIPLY(z3, - FIX_0_765366865);
    
    tmp10 = tmp0 + tmp2;
    tmp12 = tmp0 - tmp2;
    
    /* Odd part */
    
    z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
    z2 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
    z3 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
    z4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
    
    tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
	 + MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
	 + MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
	 + MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
    
    tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
	 + MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
	 + MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
	 + MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */

    /* Final output stage */
    
    wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp2, CONST_BITS-PASS1_BITS+1);
    wsptr[DCTSIZE*3] = (int) DESCALE(tmp10 - tmp2, CONST_BITS-PASS1_BITS+1);
    wsptr[DCTSIZE*1] = (int) DESCALE(tmp12 + tmp0, CONST_BITS-PASS1_BITS+1);
    wsptr[DCTSIZE*2] = (int) DESCALE(tmp12 - tmp0, CONST_BITS-PASS1_BITS+1);
  }
  
  /* Pass 2: process 4 rows from work array, store into output array. */

  wsptr = workspace;
  for (ctr = 0; ctr < 4; ctr++) {
    outptr = output_buf[ctr] + output_col;
    /* It's not clear whether a zero row test is worthwhile here ... */

#ifndef NO_ZERO_ROW_TEST
    if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 &&
	wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
      /* AC terms all zero */
      JSAMPLE dcval = range_limit[(int) DESCALE((INT32) wsptr[0], PASS1_BITS+3)
				  & RANGE_MASK];
      
      outptr[0] = dcval;
      outptr[1] = dcval;
      outptr[2] = dcval;
      outptr[3] = dcval;
      
      wsptr += DCTSIZE;		/* advance pointer to next row */
      continue;
    }
#endif
    
    /* Even part */
    
    tmp0 = ((INT32) wsptr[0]) << (CONST_BITS+1);
    
    tmp2 = MULTIPLY((INT32) wsptr[2], FIX_1_847759065)
	 + MULTIPLY((INT32) wsptr[6], - FIX_0_765366865);
    
    tmp10 = tmp0 + tmp2;
    tmp12 = tmp0 - tmp2;
    
    /* Odd part */
    
    z1 = (INT32) wsptr[7];
    z2 = (INT32) wsptr[5];
    z3 = (INT32) wsptr[3];
    z4 = (INT32) wsptr[1];
    
    tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
	 + MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
	 + MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
	 + MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
    
    tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
	 + MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
	 + MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
	 + MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */

    /* Final output stage */
    
    outptr[0] = range_limit[(int) DESCALE(tmp10 + tmp2,
					  CONST_BITS+PASS1_BITS+3+1)
			    & RANGE_MASK];
    outptr[3] = range_limit[(int) DESCALE(tmp10 - tmp2,
					  CONST_BITS+PASS1_BITS+3+1)
			    & RANGE_MASK];
    outptr[1] = range_limit[(int) DESCALE(tmp12 + tmp0,
					  CONST_BITS+PASS1_BITS+3+1)
			    & RANGE_MASK];
    outptr[2] = range_limit[(int) DESCALE(tmp12 - tmp0,
					  CONST_BITS+PASS1_BITS+3+1)
			    & RANGE_MASK];
    
    wsptr += DCTSIZE;		/* advance pointer to next row */
  }
}
Пример #3
0
void jpeg_idct_ifast (
  BLOCK_TYPE *inptr,
  short *quantptr,
  uchar * *outptr,
  int output_col)
{
  short workspace[64 + 4];	/* buffers data between passes */
  short *wsptr=workspace;

  __asm{

	mov		edi, quantptr
	mov		ebx, inptr
	mov		esi, wsptr
	add		esi, 0x07		;align wsptr to qword
	and		esi, 0xfffffff8	;align wsptr to qword

	mov		eax, esi

    /* Odd part */


	movq		mm1, [ebx + 8*10]		;load inptr[DCTSIZE*5]

	pmullw		mm1, [edi + 8*10]		;tmp6 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);

	movq		mm0, [ebx + 8*6]		;load inptr[DCTSIZE*3]

	pmullw		mm0, [edi + 8*6]		;tmp5 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);

	movq		mm3, [ebx + 8*2]		;load inptr[DCTSIZE*1]
	movq	mm2, mm1					;copy tmp6	/* phase 6 */

	pmullw		mm3, [edi + 8*2]		;tmp4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);

	movq		mm4, [ebx + 8*14]		;load inptr[DCTSIZE*1]
	paddw	mm1, mm0					;z13 = tmp6 + tmp5;

	pmullw		mm4, [edi + 8*14]	    ;tmp7 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
	psubw	mm2, mm0					;z10 = tmp6 - tmp5

	psllw		mm2, 2				;shift z10
	movq		mm0, mm2			;copy z10

	pmulhw		mm2, fix_184n261	;MULTIPLY( z12, FIX_1_847759065); /* 2*c2 */
	movq		mm5, mm3				;copy tmp4

	pmulhw		mm0, fix_n184		;MULTIPLY(z10, -FIX_1_847759065); /* 2*c2 */
	paddw		mm3, mm4				;z11 = tmp4 + tmp7;

	movq		mm6, mm3				;copy z11			/* phase 5 */
	psubw		mm5, mm4				;z12 = tmp4 - tmp7;

	psubw		mm6, mm1				;z11-z13
	psllw		mm5, 2				;shift z12

	movq		mm4, [ebx + 8*12]		;load inptr[DCTSIZE*6], even part
 	movq		mm7, mm5			;copy z12

	pmulhw		mm5, fix_108n184	;MULT(z12, (FIX_1_08-FIX_1_84)) //- z5; /* 2*(c2-c6) */ even part
	paddw		mm3, mm1				;tmp7 = z11 + z13;


    /* Even part */
	pmulhw		mm7, fix_184		;MULTIPLY(z10,(FIX_1_847759065 - FIX_2_613125930)) //+ z5; /* -2*(c2+c6) */
	psllw		mm6, 2

	movq		mm1, [ebx + 8*4]		;load inptr[DCTSIZE*2]

	pmullw		mm1, [edi + 8*4]		;tmp1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
	paddw		mm0, mm5			;tmp10

	pmullw		mm4, [edi + 8*12]		;tmp3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
	paddw		mm2, mm7			;tmp12

	pmulhw		mm6, fix_141			;tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */
	psubw		mm2, mm3		;tmp6 = tmp12 - tmp7

	movq		mm5, mm1				;copy tmp1
	paddw		mm1, mm4				;tmp13= tmp1 + tmp3;	/* phases 5-3 */

	psubw		mm5, mm4				;tmp1-tmp3
	psubw		mm6, mm2		;tmp5 = tmp11 - tmp6;

	movq		[esi+8*0], mm1			;save tmp13 in workspace
	psllw		mm5, 2					;shift tmp1-tmp3

	movq		mm7, [ebx + 8*0]		;load inptr[DCTSIZE*0]

	pmulhw		mm5, fix_141			;MULTIPLY(tmp1 - tmp3, FIX_1_414213562)
	paddw		mm0, mm6		;tmp4 = tmp10 + tmp5;

	pmullw		mm7, [edi + 8*0]		;tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);

	movq		mm4, [ebx + 8*8]		;load inptr[DCTSIZE*4]

	pmullw		mm4, [edi + 8*8]		;tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
	psubw		mm5, mm1				;tmp12 = MULTIPLY(tmp1 - tmp3, FIX_1_414213562) - tmp13; /* 2*c4 */

	movq		[esi+8*4], mm0		;save tmp4 in workspace
	movq		mm1, mm7			;copy tmp0	/* phase 3 */

	movq		[esi+8*2], mm5		;save tmp12 in workspace
	psubw		mm1, mm4			;tmp11 = tmp0 - tmp2;

	paddw		mm7, mm4			;tmp10 = tmp0 + tmp2;
    movq		mm5, mm1		;copy tmp11

	paddw		mm1, [esi+8*2]	;tmp1 = tmp11 + tmp12;
	movq		mm4, mm7		;copy tmp10		/* phase 2 */

	paddw		mm7, [esi+8*0]	;tmp0 = tmp10 + tmp13;

	psubw		mm4, [esi+8*0]	;tmp3 = tmp10 - tmp13;
	movq		mm0, mm7		;copy tmp0

	psubw		mm5, [esi+8*2]	;tmp2 = tmp11 - tmp12;
	paddw		mm7, mm3		;wsptr[DCTSIZE*0] = (int) (tmp0 + tmp7);

	psubw		mm0, mm3			;wsptr[DCTSIZE*7] = (int) (tmp0 - tmp7);

	movq		[esi + 8*0], mm7	;wsptr[DCTSIZE*0]
	movq		mm3, mm1			;copy tmp1

	movq		[esi + 8*14], mm0	;wsptr[DCTSIZE*7]
	paddw		mm1, mm2			;wsptr[DCTSIZE*1] = (int) (tmp1 + tmp6);

	psubw		mm3, mm2			;wsptr[DCTSIZE*6] = (int) (tmp1 - tmp6);

	movq		[esi + 8*2], mm1	;wsptr[DCTSIZE*1]
	movq		mm1, mm4			;copy tmp3

	movq		[esi + 8*12], mm3	;wsptr[DCTSIZE*6]

	paddw		mm4, [esi+8*4]		;wsptr[DCTSIZE*4] = (int) (tmp3 + tmp4);

	psubw		mm1, [esi+8*4]		;wsptr[DCTSIZE*3] = (int) (tmp3 - tmp4);

	movq		[esi + 8*8], mm4
	movq		mm7, mm5			;copy tmp2

	paddw		mm5, mm6			;wsptr[DCTSIZE*2] = (int) (tmp2 + tmp5)

	movq		[esi+8*6], mm1		;
	psubw		mm7, mm6			;wsptr[DCTSIZE*5] = (int) (tmp2 - tmp5);

	movq		[esi + 8*4], mm5

	movq		[esi + 8*10], mm7



/*****************************************************************/
	add		edi, 8
	add		ebx, 8
	add		esi, 8

/*****************************************************************/




	movq		mm1, [ebx + 8*10]		;load inptr[DCTSIZE*5]

	pmullw		mm1, [edi + 8*10]		;tmp6 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);

	movq		mm0, [ebx + 8*6]		;load inptr[DCTSIZE*3]

	pmullw		mm0, [edi + 8*6]		;tmp5 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);

	movq		mm3, [ebx + 8*2]		;load inptr[DCTSIZE*1]
	movq	mm2, mm1					;copy tmp6	/* phase 6 */

	pmullw		mm3, [edi + 8*2]		;tmp4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);

	movq		mm4, [ebx + 8*14]		;load inptr[DCTSIZE*1]
	paddw	mm1, mm0					;z13 = tmp6 + tmp5;

	pmullw		mm4, [edi + 8*14]	    ;tmp7 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
	psubw	mm2, mm0					;z10 = tmp6 - tmp5

	psllw		mm2, 2				;shift z10
	movq		mm0, mm2			;copy z10

	pmulhw		mm2, fix_184n261	;MULTIPLY( z12, FIX_1_847759065); /* 2*c2 */
	movq		mm5, mm3				;copy tmp4

	pmulhw		mm0, fix_n184		;MULTIPLY(z10, -FIX_1_847759065); /* 2*c2 */
	paddw		mm3, mm4				;z11 = tmp4 + tmp7;

	movq		mm6, mm3				;copy z11			/* phase 5 */
	psubw		mm5, mm4				;z12 = tmp4 - tmp7;

	psubw		mm6, mm1				;z11-z13
	psllw		mm5, 2				;shift z12

	movq		mm4, [ebx + 8*12]		;load inptr[DCTSIZE*6], even part
 	movq		mm7, mm5			;copy z12

	pmulhw		mm5, fix_108n184	;MULT(z12, (FIX_1_08-FIX_1_84)) //- z5; /* 2*(c2-c6) */ even part
	paddw		mm3, mm1				;tmp7 = z11 + z13;


    /* Even part */
	pmulhw		mm7, fix_184		;MULTIPLY(z10,(FIX_1_847759065 - FIX_2_613125930)) //+ z5; /* -2*(c2+c6) */
	psllw		mm6, 2

	movq		mm1, [ebx + 8*4]		;load inptr[DCTSIZE*2]

	pmullw		mm1, [edi + 8*4]		;tmp1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
	paddw		mm0, mm5			;tmp10

	pmullw		mm4, [edi + 8*12]		;tmp3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
	paddw		mm2, mm7			;tmp12

	pmulhw		mm6, fix_141			;tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */
	psubw		mm2, mm3		;tmp6 = tmp12 - tmp7

	movq		mm5, mm1				;copy tmp1
	paddw		mm1, mm4				;tmp13= tmp1 + tmp3;	/* phases 5-3 */

	psubw		mm5, mm4				;tmp1-tmp3
	psubw		mm6, mm2		;tmp5 = tmp11 - tmp6;

	movq		[esi+8*0], mm1			;save tmp13 in workspace
	psllw		mm5, 2					;shift tmp1-tmp3

	movq		mm7, [ebx + 8*0]		;load inptr[DCTSIZE*0]
	paddw		mm0, mm6		;tmp4 = tmp10 + tmp5;

	pmulhw		mm5, fix_141			;MULTIPLY(tmp1 - tmp3, FIX_1_414213562)

	pmullw		mm7, [edi + 8*0]		;tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);

	movq		mm4, [ebx + 8*8]		;load inptr[DCTSIZE*4]

	pmullw		mm4, [edi + 8*8]		;tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
	psubw		mm5, mm1				;tmp12 = MULTIPLY(tmp1 - tmp3, FIX_1_414213562) - tmp13; /* 2*c4 */

	movq		[esi+8*4], mm0		;save tmp4 in workspace
	movq		mm1, mm7			;copy tmp0	/* phase 3 */

	movq		[esi+8*2], mm5		;save tmp12 in workspace
	psubw		mm1, mm4			;tmp11 = tmp0 - tmp2;

	paddw		mm7, mm4			;tmp10 = tmp0 + tmp2;
    movq		mm5, mm1		;copy tmp11

	paddw		mm1, [esi+8*2]	;tmp1 = tmp11 + tmp12;
	movq		mm4, mm7		;copy tmp10		/* phase 2 */

	paddw		mm7, [esi+8*0]	;tmp0 = tmp10 + tmp13;

	psubw		mm4, [esi+8*0]	;tmp3 = tmp10 - tmp13;
	movq		mm0, mm7		;copy tmp0

	psubw		mm5, [esi+8*2]	;tmp2 = tmp11 - tmp12;
	paddw		mm7, mm3		;wsptr[DCTSIZE*0] = (int) (tmp0 + tmp7);

	psubw		mm0, mm3			;wsptr[DCTSIZE*7] = (int) (tmp0 - tmp7);

	movq		[esi + 8*0], mm7	;wsptr[DCTSIZE*0]
	movq		mm3, mm1			;copy tmp1

	movq		[esi + 8*14], mm0	;wsptr[DCTSIZE*7]
	paddw		mm1, mm2			;wsptr[DCTSIZE*1] = (int) (tmp1 + tmp6);

	psubw		mm3, mm2			;wsptr[DCTSIZE*6] = (int) (tmp1 - tmp6);

	movq		[esi + 8*2], mm1	;wsptr[DCTSIZE*1]
	movq		mm1, mm4			;copy tmp3

	movq		[esi + 8*12], mm3	;wsptr[DCTSIZE*6]

	paddw		mm4, [esi+8*4]		;wsptr[DCTSIZE*4] = (int) (tmp3 + tmp4);

	psubw		mm1, [esi+8*4]		;wsptr[DCTSIZE*3] = (int) (tmp3 - tmp4);

	movq		[esi + 8*8], mm4
	movq		mm7, mm5			;copy tmp2

	paddw		mm5, mm6			;wsptr[DCTSIZE*2] = (int) (tmp2 + tmp5)

	movq		[esi+8*6], mm1		;
	psubw		mm7, mm6			;wsptr[DCTSIZE*5] = (int) (tmp2 - tmp5);

	movq		[esi + 8*4], mm5

	movq		[esi + 8*10], mm7




/*****************************************************************/

  /* Pass 2: process rows from work array, store into output array. */
  /* Note that we must descale the results by a factor of 8 == 2**3, */
  /* and also undo the PASS1_BITS scaling. */

/*****************************************************************/
    /* Even part */

	mov			esi, eax
	mov			eax, outptr

//    tmp10 = ((DCTELEM) wsptr[0] + (DCTELEM) wsptr[4]);
//    tmp13 = ((DCTELEM) wsptr[2] + (DCTELEM) wsptr[6]);
//    tmp11 = ((DCTELEM) wsptr[0] - (DCTELEM) wsptr[4]);
//    tmp14 = ((DCTELEM) wsptr[2] - (DCTELEM) wsptr[6]);
	movq		mm0, [esi+8*0]		;wsptr[0,0],[0,1],[0,2],[0,3]

	movq		mm1, [esi+8*1]		;wsptr[0,4],[0,5],[0,6],[0,7]
	movq		mm2, mm0

	movq		mm3, [esi+8*2]		;wsptr[1,0],[1,1],[1,2],[1,3]
	paddw		mm0, mm1			;wsptr[0,tmp10],[xxx],[0,tmp13],[xxx]

	movq		mm4, [esi+8*3]		;wsptr[1,4],[1,5],[1,6],[1,7]
	psubw		mm2, mm1			;wsptr[0,tmp11],[xxx],[0,tmp14],[xxx]

	movq		mm6, mm0
	movq		mm5, mm3

	paddw		mm3, mm4			;wsptr[1,tmp10],[xxx],[1,tmp13],[xxx]
	movq		mm1, mm2

	psubw		mm5, mm4			;wsptr[1,tmp11],[xxx],[1,tmp14],[xxx]
	punpcklwd	mm0, mm3			;wsptr[0,tmp10],[1,tmp10],[xxx],[xxx]

	movq		mm7, [esi+8*7]		;wsptr[3,4],[3,5],[3,6],[3,7]
	punpckhwd	mm6, mm3			;wsptr[0,tmp13],[1,tmp13],[xxx],[xxx]

	movq		mm3, [esi+8*4]		;wsptr[2,0],[2,1],[2,2],[2,3]
	punpckldq	mm0, mm6	;wsptr[0,tmp10],[1,tmp10],[0,tmp13],[1,tmp13]

	punpcklwd	mm1, mm5			;wsptr[0,tmp11],[1,tmp11],[xxx],[xxx]
	movq		mm4, mm3

	movq		mm6, [esi+8*6]		;wsptr[3,0],[3,1],[3,2],[3,3]
	punpckhwd	mm2, mm5			;wsptr[0,tmp14],[1,tmp14],[xxx],[xxx]

	movq		mm5, [esi+8*5]		;wsptr[2,4],[2,5],[2,6],[2,7]
	punpckldq	mm1, mm2	;wsptr[0,tmp11],[1,tmp11],[0,tmp14],[1,tmp14]


	paddw		mm3, mm5			;wsptr[2,tmp10],[xxx],[2,tmp13],[xxx]
	movq		mm2, mm6

	psubw		mm4, mm5			;wsptr[2,tmp11],[xxx],[2,tmp14],[xxx]
	paddw		mm6, mm7			;wsptr[3,tmp10],[xxx],[3,tmp13],[xxx]

	movq		mm5, mm3
	punpcklwd	mm3, mm6			;wsptr[2,tmp10],[3,tmp10],[xxx],[xxx]

	psubw		mm2, mm7			;wsptr[3,tmp11],[xxx],[3,tmp14],[xxx]
	punpckhwd	mm5, mm6			;wsptr[2,tmp13],[3,tmp13],[xxx],[xxx]

	movq		mm7, mm4
	punpckldq	mm3, mm5	;wsptr[2,tmp10],[3,tmp10],[2,tmp13],[3,tmp13]

	punpcklwd	mm4, mm2			;wsptr[2,tmp11],[3,tmp11],[xxx],[xxx]

	punpckhwd	mm7, mm2			;wsptr[2,tmp14],[3,tmp14],[xxx],[xxx]

	punpckldq	mm4, mm7	;wsptr[2,tmp11],[3,tmp11],[2,tmp14],[3,tmp14]
	movq		mm6, mm1

//	mm0 = 	;wsptr[0,tmp10],[1,tmp10],[0,tmp13],[1,tmp13]
//	mm1 =	;wsptr[0,tmp11],[1,tmp11],[0,tmp14],[1,tmp14]


	movq		mm2, mm0
	punpckhdq	mm6, mm4	;wsptr[0,tmp14],[1,tmp14],[2,tmp14],[3,tmp14]

	punpckldq	mm1, mm4	;wsptr[0,tmp11],[1,tmp11],[2,tmp11],[3,tmp11]
	psllw		mm6, 2

	pmulhw		mm6, fix_141
	punpckldq	mm0, mm3	;wsptr[0,tmp10],[1,tmp10],[2,tmp10],[3,tmp10]

	punpckhdq	mm2, mm3	;wsptr[0,tmp13],[1,tmp13],[2,tmp13],[3,tmp13]
	movq		mm7, mm0

//    tmp0 = tmp10 + tmp13;
//    tmp3 = tmp10 - tmp13;
	paddw		mm0, mm2	;[0,tmp0],[1,tmp0],[2,tmp0],[3,tmp0]
	psubw		mm7, mm2	;[0,tmp3],[1,tmp3],[2,tmp3],[3,tmp3]

//    tmp12 = MULTIPLY(tmp14, FIX_1_414213562) - tmp13;
	psubw		mm6, mm2	;wsptr[0,tmp12],[1,tmp12],[2,tmp12],[3,tmp12]
//    tmp1 = tmp11 + tmp12;
//    tmp2 = tmp11 - tmp12;
	movq		mm5, mm1



    /* Odd part */

//    z13 = (DCTELEM) wsptr[5] + (DCTELEM) wsptr[3];
//    z10 = (DCTELEM) wsptr[5] - (DCTELEM) wsptr[3];
//    z11 = (DCTELEM) wsptr[1] + (DCTELEM) wsptr[7];
//    z12 = (DCTELEM) wsptr[1] - (DCTELEM) wsptr[7];
	movq		mm3, [esi+8*0]		;wsptr[0,0],[0,1],[0,2],[0,3]
	paddw		mm1, mm6	;[0,tmp1],[1,tmp1],[2,tmp1],[3,tmp1]

	movq		mm4, [esi+8*1]		;wsptr[0,4],[0,5],[0,6],[0,7]
	psubw		mm5, mm6	;[0,tmp2],[1,tmp2],[2,tmp2],[3,tmp2]

	movq		mm6, mm3
	punpckldq	mm3, mm4			;wsptr[0,0],[0,1],[0,4],[0,5]

	punpckhdq	mm4, mm6			;wsptr[0,6],[0,7],[0,2],[0,3]
	movq		mm2, mm3

//Save tmp0 and tmp1 in wsptr
	movq		[esi+8*0], mm0		;save tmp0
	paddw		mm2, mm4			;wsptr[xxx],[0,z11],[xxx],[0,z13]


//Continue with z10 --- z13
	movq		mm6, [esi+8*2]		;wsptr[1,0],[1,1],[1,2],[1,3]
	psubw		mm3, mm4			;wsptr[xxx],[0,z12],[xxx],[0,z10]

	movq		mm0, [esi+8*3]		;wsptr[1,4],[1,5],[1,6],[1,7]
	movq		mm4, mm6

	movq		[esi+8*1], mm1		;save tmp1
	punpckldq	mm6, mm0			;wsptr[1,0],[1,1],[1,4],[1,5]

	punpckhdq	mm0, mm4			;wsptr[1,6],[1,7],[1,2],[1,3]
	movq		mm1, mm6

//Save tmp2 and tmp3 in wsptr
	paddw		mm6, mm0		;wsptr[xxx],[1,z11],[xxx],[1,z13]
	movq		mm4, mm2

//Continue with z10 --- z13
	movq		[esi+8*2], mm5		;save tmp2
	punpcklwd	mm2, mm6		;wsptr[xxx],[xxx],[0,z11],[1,z11]

	psubw		mm1, mm0		;wsptr[xxx],[1,z12],[xxx],[1,z10]
	punpckhwd	mm4, mm6		;wsptr[xxx],[xxx],[0,z13],[1,z13]

	movq		mm0, mm3
	punpcklwd	mm3, mm1		;wsptr[xxx],[xxx],[0,z12],[1,z12]

	movq		[esi+8*3], mm7		;save tmp3
	punpckhwd	mm0, mm1		;wsptr[xxx],[xxx],[0,z10],[1,z10]

	movq		mm6, [esi+8*4]		;wsptr[2,0],[2,1],[2,2],[2,3]
	punpckhdq	mm0, mm2		;wsptr[0,z10],[1,z10],[0,z11],[1,z11]

	movq		mm7, [esi+8*5]		;wsptr[2,4],[2,5],[2,6],[2,7]
	punpckhdq	mm3, mm4		;wsptr[0,z12],[1,z12],[0,z13],[1,z13]

	movq		mm1, [esi+8*6]		;wsptr[3,0],[3,1],[3,2],[3,3]
	movq		mm4, mm6

	punpckldq	mm6, mm7			;wsptr[2,0],[2,1],[2,4],[2,5]
	movq		mm5, mm1

	punpckhdq	mm7, mm4			;wsptr[2,6],[2,7],[2,2],[2,3]
	movq		mm2, mm6

	movq		mm4, [esi+8*7]		;wsptr[3,4],[3,5],[3,6],[3,7]
	paddw		mm6, mm7		;wsptr[xxx],[2,z11],[xxx],[2,z13]

	psubw		mm2, mm7		;wsptr[xxx],[2,z12],[xxx],[2,z10]
	punpckldq	mm1, mm4			;wsptr[3,0],[3,1],[3,4],[3,5]

	punpckhdq	mm4, mm5			;wsptr[3,6],[3,7],[3,2],[3,3]
	movq		mm7, mm1

	paddw		mm1, mm4		;wsptr[xxx],[3,z11],[xxx],[3,z13]
	psubw		mm7, mm4		;wsptr[xxx],[3,z12],[xxx],[3,z10]

	movq		mm5, mm6
	punpcklwd	mm6, mm1		;wsptr[xxx],[xxx],[2,z11],[3,z11]

	punpckhwd	mm5, mm1		;wsptr[xxx],[xxx],[2,z13],[3,z13]
	movq		mm4, mm2

	punpcklwd	mm2, mm7		;wsptr[xxx],[xxx],[2,z12],[3,z12]

	punpckhwd	mm4, mm7		;wsptr[xxx],[xxx],[2,z10],[3,z10]

	punpckhdq	mm4, mm6		;wsptr[2,z10],[3,z10],[2,z11],[3,z11]

	punpckhdq	mm2, mm5		;wsptr[2,z12],[3,z12],[2,z13],[3,z13]
	movq		mm5, mm0

	punpckldq	mm0, mm4		;wsptr[0,z10],[1,z10],[2,z10],[3,z10]

	punpckhdq	mm5, mm4		;wsptr[0,z11],[1,z11],[2,z11],[3,z11]
	movq		mm4, mm3

	punpckhdq	mm4, mm2		;wsptr[0,z13],[1,z13],[2,z13],[3,z13]
	movq		mm1, mm5

	punpckldq	mm3, mm2		;wsptr[0,z12],[1,z12],[2,z12],[3,z12]
//    tmp7 = z11 + z13;		/* phase 5 */
//    tmp8 = z11 - z13;		/* phase 5 */
	psubw		mm1, mm4		;tmp8

	paddw		mm5, mm4		;tmp7
//    tmp21 = MULTIPLY(tmp8, FIX_1_414213562); /* 2*c4 */
	psllw		mm1, 2

	psllw		mm0, 2

	pmulhw		mm1, fix_141	;tmp21
//    tmp20 = MULTIPLY(z12, (FIX_1_082392200- FIX_1_847759065))  /* 2*(c2-c6) */
//			+ MULTIPLY(z10, - FIX_1_847759065); /* 2*c2 */
	psllw		mm3, 2
	movq		mm7, mm0

	pmulhw		mm7, fix_n184
	movq		mm6, mm3

	movq		mm2, [esi+8*0]	;tmp0,final1

	pmulhw		mm6, fix_108n184
//	 tmp22 = MULTIPLY(z10,(FIX_1_847759065 - FIX_2_613125930)) /* -2*(c2+c6) */
//			+ MULTIPLY(z12, FIX_1_847759065); /* 2*c2 */
	movq		mm4, mm2		;final1

	pmulhw		mm0, fix_184n261
	paddw		mm2, mm5		;tmp0+tmp7,final1

	pmulhw		mm3, fix_184
	psubw		mm4, mm5		;tmp0-tmp7,final1

//    tmp6 = tmp22 - tmp7;	/* phase 2 */
	psraw		mm2, 5			;outptr[0,0],[1,0],[2,0],[3,0],final1

	paddsw		mm2, const_0x0080	;final1
	paddw		mm7, mm6			;tmp20
	psraw		mm4, 5			;outptr[0,7],[1,7],[2,7],[3,7],final1

	paddsw		mm4, const_0x0080	;final1
	paddw		mm3, mm0			;tmp22

//    tmp5 = tmp21 - tmp6;
	psubw		mm3, mm5		;tmp6

//    tmp4 = tmp20 + tmp5;
	movq		mm0, [esi+8*1]		;tmp1,final2
	psubw		mm1, mm3		;tmp5

	movq		mm6, mm0			;final2
	paddw		mm0, mm3		;tmp1+tmp6,final2

    /* Final output stage: scale down by a factor of 8 and range-limit */


//    outptr[0] = range_limit[IDESCALE(tmp0 + tmp7, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[7] = range_limit[IDESCALE(tmp0 - tmp7, PASS1_BITS+3)
//			    & RANGE_MASK];	final1


//    outptr[1] = range_limit[IDESCALE(tmp1 + tmp6, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[6] = range_limit[IDESCALE(tmp1 - tmp6, PASS1_BITS+3)
//			    & RANGE_MASK];	final2
	psubw		mm6, mm3		;tmp1-tmp6,final2
	psraw		mm0, 5			;outptr[0,1],[1,1],[2,1],[3,1]

	paddsw		mm0, const_0x0080
	psraw		mm6, 5			;outptr[0,6],[1,6],[2,6],[3,6]

	paddsw		mm6, const_0x0080		;need to check this value
	packuswb	mm0, mm4	;out[0,1],[1,1],[2,1],[3,1],[0,7],[1,7],[2,7],[3,7]

	movq		mm5, [esi+8*2]		;tmp2,final3
	packuswb	mm2, mm6	;out[0,0],[1,0],[2,0],[3,0],[0,6],[1,6],[2,6],[3,6]

//    outptr[2] = range_limit[IDESCALE(tmp2 + tmp5, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[5] = range_limit[IDESCALE(tmp2 - tmp5, PASS1_BITS+3)
//			    & RANGE_MASK];	final3
	paddw		mm7, mm1		;tmp4
	movq		mm3, mm5

	paddw		mm5, mm1		;tmp2+tmp5
	psubw		mm3, mm1		;tmp2-tmp5

	psraw		mm5, 5			;outptr[0,2],[1,2],[2,2],[3,2]

	paddsw		mm5, const_0x0080
	movq		mm4, [esi+8*3]		;tmp3,final4
	psraw		mm3, 5			;outptr[0,5],[1,5],[2,5],[3,5]

	paddsw		mm3, const_0x0080


//    outptr[4] = range_limit[IDESCALE(tmp3 + tmp4, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[3] = range_limit[IDESCALE(tmp3 - tmp4, PASS1_BITS+3)
//			    & RANGE_MASK];	final4
	movq		mm6, mm4
	paddw		mm4, mm7		;tmp3+tmp4

	psubw		mm6, mm7		;tmp3-tmp4
	psraw		mm4, 5			;outptr[0,4],[1,4],[2,4],[3,4]
	mov			ecx, [eax]

	paddsw		mm4, const_0x0080
	psraw		mm6, 5			;outptr[0,3],[1,3],[2,3],[3,3]

	paddsw		mm6, const_0x0080
	packuswb	mm5, mm4	;out[0,2],[1,2],[2,2],[3,2],[0,4],[1,4],[2,4],[3,4]

	packuswb	mm6, mm3	;out[0,3],[1,3],[2,3],[3,3],[0,5],[1,5],[2,5],[3,5]
	movq		mm4, mm2

	movq		mm7, mm5
	punpcklbw	mm2, mm0	;out[0,0],[0,1],[1,0],[1,1],[2,0],[2,1],[3,0],[3,1]

	punpckhbw	mm4, mm0	;out[0,6],[0,7],[1,6],[1,7],[2,6],[2,7],[3,6],[3,7]
	movq		mm1, mm2

	punpcklbw	mm5, mm6	;out[0,2],[0,3],[1,2],[1,3],[2,2],[2,3],[3,2],[3,3]
	add		 	eax, 4

	punpckhbw	mm7, mm6	;out[0,4],[0,5],[1,4],[1,5],[2,4],[2,5],[3,4],[3,5]

	punpcklwd	mm2, mm5	;out[0,0],[0,1],[0,2],[0,3],[1,0],[1,1],[1,2],[1,3]
	add			ecx, output_col

	movq		mm6, mm7
	punpckhwd	mm1, mm5	;out[2,0],[2,1],[2,2],[2,3],[3,0],[3,1],[3,2],[3,3]

	movq		mm0, mm2
	punpcklwd	mm6, mm4	;out[0,4],[0,5],[0,6],[0,7],[1,4],[1,5],[1,6],[1,7]

	mov			ebx, [eax]
	punpckldq	mm2, mm6	;out[0,0],[0,1],[0,2],[0,3],[0,4],[0,5],[0,6],[0,7]

	add		 	eax, 4
	movq		mm3, mm1

	add			ebx, output_col
	punpckhwd	mm7, mm4	;out[2,4],[2,5],[2,6],[2,7],[3,4],[3,5],[3,6],[3,7]

	movq		[ecx], mm2
	punpckhdq	mm0, mm6	;out[1,0],[1,1],[1,2],[1,3],[1,4],[1,5],[1,6],[1,7]

	mov			ecx, [eax]
	add		 	eax, 4
	add			ecx, output_col

	movq		[ebx], mm0
	punpckldq	mm1, mm7	;out[2,0],[2,1],[2,2],[2,3],[2,4],[2,5],[2,6],[2,7]

	mov			ebx, [eax]

	add			ebx, output_col
	punpckhdq	mm3, mm7	;out[3,0],[3,1],[3,2],[3,3],[3,4],[3,5],[3,6],[3,7]
	movq		[ecx], mm1


	movq		[ebx], mm3



/*******************************************************************/


	add			esi, 64
	add			eax, 4

/*******************************************************************/

//    tmp10 = ((DCTELEM) wsptr[0] + (DCTELEM) wsptr[4]);
//    tmp13 = ((DCTELEM) wsptr[2] + (DCTELEM) wsptr[6]);
//    tmp11 = ((DCTELEM) wsptr[0] - (DCTELEM) wsptr[4]);
//    tmp14 = ((DCTELEM) wsptr[2] - (DCTELEM) wsptr[6]);
	movq		mm0, [esi+8*0]		;wsptr[0,0],[0,1],[0,2],[0,3]

	movq		mm1, [esi+8*1]		;wsptr[0,4],[0,5],[0,6],[0,7]
	movq		mm2, mm0

	movq		mm3, [esi+8*2]		;wsptr[1,0],[1,1],[1,2],[1,3]
	paddw		mm0, mm1			;wsptr[0,tmp10],[xxx],[0,tmp13],[xxx]

	movq		mm4, [esi+8*3]		;wsptr[1,4],[1,5],[1,6],[1,7]
	psubw		mm2, mm1			;wsptr[0,tmp11],[xxx],[0,tmp14],[xxx]

	movq		mm6, mm0
	movq		mm5, mm3

	paddw		mm3, mm4			;wsptr[1,tmp10],[xxx],[1,tmp13],[xxx]
	movq		mm1, mm2

	psubw		mm5, mm4			;wsptr[1,tmp11],[xxx],[1,tmp14],[xxx]
	punpcklwd	mm0, mm3			;wsptr[0,tmp10],[1,tmp10],[xxx],[xxx]

	movq		mm7, [esi+8*7]		;wsptr[3,4],[3,5],[3,6],[3,7]
	punpckhwd	mm6, mm3			;wsptr[0,tmp13],[1,tmp13],[xxx],[xxx]

	movq		mm3, [esi+8*4]		;wsptr[2,0],[2,1],[2,2],[2,3]
	punpckldq	mm0, mm6	;wsptr[0,tmp10],[1,tmp10],[0,tmp13],[1,tmp13]

	punpcklwd	mm1, mm5			;wsptr[0,tmp11],[1,tmp11],[xxx],[xxx]
	movq		mm4, mm3

	movq		mm6, [esi+8*6]		;wsptr[3,0],[3,1],[3,2],[3,3]
	punpckhwd	mm2, mm5			;wsptr[0,tmp14],[1,tmp14],[xxx],[xxx]

	movq		mm5, [esi+8*5]		;wsptr[2,4],[2,5],[2,6],[2,7]
	punpckldq	mm1, mm2	;wsptr[0,tmp11],[1,tmp11],[0,tmp14],[1,tmp14]


	paddw		mm3, mm5			;wsptr[2,tmp10],[xxx],[2,tmp13],[xxx]
	movq		mm2, mm6

	psubw		mm4, mm5			;wsptr[2,tmp11],[xxx],[2,tmp14],[xxx]
	paddw		mm6, mm7			;wsptr[3,tmp10],[xxx],[3,tmp13],[xxx]

	movq		mm5, mm3
	punpcklwd	mm3, mm6			;wsptr[2,tmp10],[3,tmp10],[xxx],[xxx]

	psubw		mm2, mm7			;wsptr[3,tmp11],[xxx],[3,tmp14],[xxx]
	punpckhwd	mm5, mm6			;wsptr[2,tmp13],[3,tmp13],[xxx],[xxx]

	movq		mm7, mm4
	punpckldq	mm3, mm5	;wsptr[2,tmp10],[3,tmp10],[2,tmp13],[3,tmp13]

	punpcklwd	mm4, mm2			;wsptr[2,tmp11],[3,tmp11],[xxx],[xxx]

	punpckhwd	mm7, mm2			;wsptr[2,tmp14],[3,tmp14],[xxx],[xxx]

	punpckldq	mm4, mm7	;wsptr[2,tmp11],[3,tmp11],[2,tmp14],[3,tmp14]
	movq		mm6, mm1

//	mm0 = 	;wsptr[0,tmp10],[1,tmp10],[0,tmp13],[1,tmp13]
//	mm1 =	;wsptr[0,tmp11],[1,tmp11],[0,tmp14],[1,tmp14]


	movq		mm2, mm0
	punpckhdq	mm6, mm4	;wsptr[0,tmp14],[1,tmp14],[2,tmp14],[3,tmp14]

	punpckldq	mm1, mm4	;wsptr[0,tmp11],[1,tmp11],[2,tmp11],[3,tmp11]
	psllw		mm6, 2

	pmulhw		mm6, fix_141
	punpckldq	mm0, mm3	;wsptr[0,tmp10],[1,tmp10],[2,tmp10],[3,tmp10]

	punpckhdq	mm2, mm3	;wsptr[0,tmp13],[1,tmp13],[2,tmp13],[3,tmp13]
	movq		mm7, mm0

//    tmp0 = tmp10 + tmp13;
//    tmp3 = tmp10 - tmp13;
	paddw		mm0, mm2	;[0,tmp0],[1,tmp0],[2,tmp0],[3,tmp0]
	psubw		mm7, mm2	;[0,tmp3],[1,tmp3],[2,tmp3],[3,tmp3]

//    tmp12 = MULTIPLY(tmp14, FIX_1_414213562) - tmp13;
	psubw		mm6, mm2	;wsptr[0,tmp12],[1,tmp12],[2,tmp12],[3,tmp12]
//    tmp1 = tmp11 + tmp12;
//    tmp2 = tmp11 - tmp12;
	movq		mm5, mm1



    /* Odd part */

//    z13 = (DCTELEM) wsptr[5] + (DCTELEM) wsptr[3];
//    z10 = (DCTELEM) wsptr[5] - (DCTELEM) wsptr[3];
//    z11 = (DCTELEM) wsptr[1] + (DCTELEM) wsptr[7];
//    z12 = (DCTELEM) wsptr[1] - (DCTELEM) wsptr[7];
	movq		mm3, [esi+8*0]		;wsptr[0,0],[0,1],[0,2],[0,3]
	paddw		mm1, mm6	;[0,tmp1],[1,tmp1],[2,tmp1],[3,tmp1]

	movq		mm4, [esi+8*1]		;wsptr[0,4],[0,5],[0,6],[0,7]
	psubw		mm5, mm6	;[0,tmp2],[1,tmp2],[2,tmp2],[3,tmp2]

	movq		mm6, mm3
	punpckldq	mm3, mm4			;wsptr[0,0],[0,1],[0,4],[0,5]

	punpckhdq	mm4, mm6			;wsptr[0,6],[0,7],[0,2],[0,3]
	movq		mm2, mm3

//Save tmp0 and tmp1 in wsptr
	movq		[esi+8*0], mm0		;save tmp0
	paddw		mm2, mm4			;wsptr[xxx],[0,z11],[xxx],[0,z13]


//Continue with z10 --- z13
	movq		mm6, [esi+8*2]		;wsptr[1,0],[1,1],[1,2],[1,3]
	psubw		mm3, mm4			;wsptr[xxx],[0,z12],[xxx],[0,z10]

	movq		mm0, [esi+8*3]		;wsptr[1,4],[1,5],[1,6],[1,7]
	movq		mm4, mm6

	movq		[esi+8*1], mm1		;save tmp1
	punpckldq	mm6, mm0			;wsptr[1,0],[1,1],[1,4],[1,5]

	punpckhdq	mm0, mm4			;wsptr[1,6],[1,7],[1,2],[1,3]
	movq		mm1, mm6

//Save tmp2 and tmp3 in wsptr
	paddw		mm6, mm0		;wsptr[xxx],[1,z11],[xxx],[1,z13]
	movq		mm4, mm2

//Continue with z10 --- z13
	movq		[esi+8*2], mm5		;save tmp2
	punpcklwd	mm2, mm6		;wsptr[xxx],[xxx],[0,z11],[1,z11]

	psubw		mm1, mm0		;wsptr[xxx],[1,z12],[xxx],[1,z10]
	punpckhwd	mm4, mm6		;wsptr[xxx],[xxx],[0,z13],[1,z13]

	movq		mm0, mm3
	punpcklwd	mm3, mm1		;wsptr[xxx],[xxx],[0,z12],[1,z12]

	movq		[esi+8*3], mm7		;save tmp3
	punpckhwd	mm0, mm1		;wsptr[xxx],[xxx],[0,z10],[1,z10]

	movq		mm6, [esi+8*4]		;wsptr[2,0],[2,1],[2,2],[2,3]
	punpckhdq	mm0, mm2		;wsptr[0,z10],[1,z10],[0,z11],[1,z11]

	movq		mm7, [esi+8*5]		;wsptr[2,4],[2,5],[2,6],[2,7]
	punpckhdq	mm3, mm4		;wsptr[0,z12],[1,z12],[0,z13],[1,z13]

	movq		mm1, [esi+8*6]		;wsptr[3,0],[3,1],[3,2],[3,3]
	movq		mm4, mm6

	punpckldq	mm6, mm7			;wsptr[2,0],[2,1],[2,4],[2,5]
	movq		mm5, mm1

	punpckhdq	mm7, mm4			;wsptr[2,6],[2,7],[2,2],[2,3]
	movq		mm2, mm6

	movq		mm4, [esi+8*7]		;wsptr[3,4],[3,5],[3,6],[3,7]
	paddw		mm6, mm7		;wsptr[xxx],[2,z11],[xxx],[2,z13]

	psubw		mm2, mm7		;wsptr[xxx],[2,z12],[xxx],[2,z10]
	punpckldq	mm1, mm4			;wsptr[3,0],[3,1],[3,4],[3,5]

	punpckhdq	mm4, mm5			;wsptr[3,6],[3,7],[3,2],[3,3]
	movq		mm7, mm1

	paddw		mm1, mm4		;wsptr[xxx],[3,z11],[xxx],[3,z13]
	psubw		mm7, mm4		;wsptr[xxx],[3,z12],[xxx],[3,z10]

	movq		mm5, mm6
	punpcklwd	mm6, mm1		;wsptr[xxx],[xxx],[2,z11],[3,z11]

	punpckhwd	mm5, mm1		;wsptr[xxx],[xxx],[2,z13],[3,z13]
	movq		mm4, mm2

	punpcklwd	mm2, mm7		;wsptr[xxx],[xxx],[2,z12],[3,z12]

	punpckhwd	mm4, mm7		;wsptr[xxx],[xxx],[2,z10],[3,z10]

	punpckhdq	mm4, mm6		;wsptr[2,z10],[3,z10],[2,z11],[3,z11]

	punpckhdq	mm2, mm5		;wsptr[2,z12],[3,z12],[2,z13],[3,z13]
	movq		mm5, mm0

	punpckldq	mm0, mm4		;wsptr[0,z10],[1,z10],[2,z10],[3,z10]

	punpckhdq	mm5, mm4		;wsptr[0,z11],[1,z11],[2,z11],[3,z11]
	movq		mm4, mm3

	punpckhdq	mm4, mm2		;wsptr[0,z13],[1,z13],[2,z13],[3,z13]
	movq		mm1, mm5

	punpckldq	mm3, mm2		;wsptr[0,z12],[1,z12],[2,z12],[3,z12]
//    tmp7 = z11 + z13;		/* phase 5 */
//    tmp8 = z11 - z13;		/* phase 5 */
	psubw		mm1, mm4		;tmp8

	paddw		mm5, mm4		;tmp7
//    tmp21 = MULTIPLY(tmp8, FIX_1_414213562); /* 2*c4 */
	psllw		mm1, 2

	psllw		mm0, 2

	pmulhw		mm1, fix_141	;tmp21
//    tmp20 = MULTIPLY(z12, (FIX_1_082392200- FIX_1_847759065))  /* 2*(c2-c6) */
//			+ MULTIPLY(z10, - FIX_1_847759065); /* 2*c2 */
	psllw		mm3, 2
	movq		mm7, mm0

	pmulhw		mm7, fix_n184
	movq		mm6, mm3

	movq		mm2, [esi+8*0]	;tmp0,final1

	pmulhw		mm6, fix_108n184
//	 tmp22 = MULTIPLY(z10,(FIX_1_847759065 - FIX_2_613125930)) /* -2*(c2+c6) */
//			+ MULTIPLY(z12, FIX_1_847759065); /* 2*c2 */
	movq		mm4, mm2		;final1

	pmulhw		mm0, fix_184n261
	paddw		mm2, mm5		;tmp0+tmp7,final1

	pmulhw		mm3, fix_184
	psubw		mm4, mm5		;tmp0-tmp7,final1

//    tmp6 = tmp22 - tmp7;	/* phase 2 */
	psraw		mm2, 5			;outptr[0,0],[1,0],[2,0],[3,0],final1

	paddsw		mm2, const_0x0080	;final1
	paddw		mm7, mm6			;tmp20
	psraw		mm4, 5			;outptr[0,7],[1,7],[2,7],[3,7],final1

	paddsw		mm4, const_0x0080	;final1
	paddw		mm3, mm0			;tmp22

//    tmp5 = tmp21 - tmp6;
	psubw		mm3, mm5		;tmp6

//    tmp4 = tmp20 + tmp5;
	movq		mm0, [esi+8*1]		;tmp1,final2
	psubw		mm1, mm3		;tmp5

	movq		mm6, mm0			;final2
	paddw		mm0, mm3		;tmp1+tmp6,final2

    /* Final output stage: scale down by a factor of 8 and range-limit */


//    outptr[0] = range_limit[IDESCALE(tmp0 + tmp7, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[7] = range_limit[IDESCALE(tmp0 - tmp7, PASS1_BITS+3)
//			    & RANGE_MASK];	final1


//    outptr[1] = range_limit[IDESCALE(tmp1 + tmp6, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[6] = range_limit[IDESCALE(tmp1 - tmp6, PASS1_BITS+3)
//			    & RANGE_MASK];	final2
	psubw		mm6, mm3		;tmp1-tmp6,final2
	psraw		mm0, 5			;outptr[0,1],[1,1],[2,1],[3,1]

	paddsw		mm0, const_0x0080
	psraw		mm6, 5			;outptr[0,6],[1,6],[2,6],[3,6]

	paddsw		mm6, const_0x0080		;need to check this value
	packuswb	mm0, mm4	;out[0,1],[1,1],[2,1],[3,1],[0,7],[1,7],[2,7],[3,7]

	movq		mm5, [esi+8*2]		;tmp2,final3
	packuswb	mm2, mm6	;out[0,0],[1,0],[2,0],[3,0],[0,6],[1,6],[2,6],[3,6]

//    outptr[2] = range_limit[IDESCALE(tmp2 + tmp5, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[5] = range_limit[IDESCALE(tmp2 - tmp5, PASS1_BITS+3)
//			    & RANGE_MASK];	final3
	paddw		mm7, mm1		;tmp4
	movq		mm3, mm5

	paddw		mm5, mm1		;tmp2+tmp5
	psubw		mm3, mm1		;tmp2-tmp5

	psraw		mm5, 5			;outptr[0,2],[1,2],[2,2],[3,2]

	paddsw		mm5, const_0x0080
	movq		mm4, [esi+8*3]		;tmp3,final4
	psraw		mm3, 5			;outptr[0,5],[1,5],[2,5],[3,5]

	paddsw		mm3, const_0x0080


//    outptr[4] = range_limit[IDESCALE(tmp3 + tmp4, PASS1_BITS+3)
//			    & RANGE_MASK];
//    outptr[3] = range_limit[IDESCALE(tmp3 - tmp4, PASS1_BITS+3)
//			    & RANGE_MASK];	final4
	movq		mm6, mm4
	paddw		mm4, mm7		;tmp3+tmp4

	psubw		mm6, mm7		;tmp3-tmp4
	psraw		mm4, 5			;outptr[0,4],[1,4],[2,4],[3,4]
	mov			ecx, [eax]

	paddsw		mm4, const_0x0080
	psraw		mm6, 5			;outptr[0,3],[1,3],[2,3],[3,3]

	paddsw		mm6, const_0x0080
	packuswb	mm5, mm4	;out[0,2],[1,2],[2,2],[3,2],[0,4],[1,4],[2,4],[3,4]

	packuswb	mm6, mm3	;out[0,3],[1,3],[2,3],[3,3],[0,5],[1,5],[2,5],[3,5]
	movq		mm4, mm2

	movq		mm7, mm5
	punpcklbw	mm2, mm0	;out[0,0],[0,1],[1,0],[1,1],[2,0],[2,1],[3,0],[3,1]

	punpckhbw	mm4, mm0	;out[0,6],[0,7],[1,6],[1,7],[2,6],[2,7],[3,6],[3,7]
	movq		mm1, mm2

	punpcklbw	mm5, mm6	;out[0,2],[0,3],[1,2],[1,3],[2,2],[2,3],[3,2],[3,3]
	add		 	eax, 4

	punpckhbw	mm7, mm6	;out[0,4],[0,5],[1,4],[1,5],[2,4],[2,5],[3,4],[3,5]

	punpcklwd	mm2, mm5	;out[0,0],[0,1],[0,2],[0,3],[1,0],[1,1],[1,2],[1,3]
	add			ecx, output_col

	movq		mm6, mm7
	punpckhwd	mm1, mm5	;out[2,0],[2,1],[2,2],[2,3],[3,0],[3,1],[3,2],[3,3]

	movq		mm0, mm2
	punpcklwd	mm6, mm4	;out[0,4],[0,5],[0,6],[0,7],[1,4],[1,5],[1,6],[1,7]

	mov			ebx, [eax]
	punpckldq	mm2, mm6	;out[0,0],[0,1],[0,2],[0,3],[0,4],[0,5],[0,6],[0,7]

	add		 	eax, 4
	movq		mm3, mm1

	add			ebx, output_col
	punpckhwd	mm7, mm4	;out[2,4],[2,5],[2,6],[2,7],[3,4],[3,5],[3,6],[3,7]

	movq		[ecx], mm2
	punpckhdq	mm0, mm6	;out[1,0],[1,1],[1,2],[1,3],[1,4],[1,5],[1,6],[1,7]

	mov			ecx, [eax]
	add		 	eax, 4
	add			ecx, output_col

	movq		[ebx], mm0
	punpckldq	mm1, mm7	;out[2,0],[2,1],[2,2],[2,3],[2,4],[2,5],[2,6],[2,7]

	mov			ebx, [eax]

	add			ebx, output_col
	punpckhdq	mm3, mm7	;out[3,0],[3,1],[3,2],[3,3],[3,4],[3,5],[3,6],[3,7]
	movq		[ecx], mm1

	movq		[ebx], mm3
	}
}
Пример #4
0
fdct_ifast (DCTELEM * data)
{
  int16_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
  int16_t tmp10, tmp11, tmp12, tmp13;
  int16_t z1, z2, z3, z4, z5, z11, z13;
  DCTELEM *dataptr;
  int ctr;
  SHIFT_TEMPS

  row_fdct(data);

  /* Pass 2: process columns. */

  dataptr = data;
  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
    tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
    tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
    tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
    tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
    tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
    tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
    tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
    tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];

    /* Even part */

    tmp10 = tmp0 + tmp3;        /* phase 2 */
    tmp13 = tmp0 - tmp3;
    tmp11 = tmp1 + tmp2;
    tmp12 = tmp1 - tmp2;

    dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */
    dataptr[DCTSIZE*4] = tmp10 - tmp11;

    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */
    dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */
    dataptr[DCTSIZE*6] = tmp13 - z1;

    /* Odd part */

    tmp10 = tmp4 + tmp5;        /* phase 2 */
    tmp11 = tmp5 + tmp6;
    tmp12 = tmp6 + tmp7;

    /* The rotator is modified from fig 4-8 to avoid extra negations. */
    z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */
    z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */
    z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */
    z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */

    z11 = tmp7 + z3;            /* phase 5 */
    z13 = tmp7 - z3;

    dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */
    dataptr[DCTSIZE*3] = z13 - z2;
    dataptr[DCTSIZE*1] = z11 + z4;
    dataptr[DCTSIZE*7] = z11 - z4;

    dataptr++;                  /* advance pointer to next column */
  }
}
Пример #5
0
jpeg_idct_2x2 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
	       JCOEFPTR coef_block,
	       JSAMPARRAY output_buf, JDIMENSION output_col)
{
  INT32 tmp0, tmp10, z1;
  JCOEFPTR inptr;
  ISLOW_MULT_TYPE * quantptr;
  int * wsptr;
  JSAMPROW outptr;
  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
  int ctr;
  int workspace[DCTSIZE*2];	/* buffers data between passes */
  SHIFT_TEMPS

  /* Pass 1: process columns from input, store into work array. */

  inptr = coef_block;
  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
  wsptr = workspace;
  for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) {
    /* Don't bother to process columns 2,4,6 */
    if (ctr == DCTSIZE-2 || ctr == DCTSIZE-4 || ctr == DCTSIZE-6)
      continue;
    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*3] == 0 &&
	inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*7] == 0) {
      /* AC terms all zero; we need not examine terms 2,4,6 for 2x2 output */
      int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
      
      wsptr[DCTSIZE*0] = dcval;
      wsptr[DCTSIZE*1] = dcval;
      
      continue;
    }
    
    /* Even part */
    
    z1 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
    tmp10 = z1 << (CONST_BITS+2);
    
    /* Odd part */

    z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
    tmp0 = MULTIPLY(z1, - FIX_0_720959822); /* sqrt(2) * (c7-c5+c3-c1) */
    z1 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
    tmp0 += MULTIPLY(z1, FIX_0_850430095); /* sqrt(2) * (-c1+c3+c5+c7) */
    z1 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
    tmp0 += MULTIPLY(z1, - FIX_1_272758580); /* sqrt(2) * (-c1+c3-c5-c7) */
    z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
    tmp0 += MULTIPLY(z1, FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */

    /* Final output stage */
    
    wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp0, CONST_BITS-PASS1_BITS+2);
    wsptr[DCTSIZE*1] = (int) DESCALE(tmp10 - tmp0, CONST_BITS-PASS1_BITS+2);
  }
  
  /* Pass 2: process 2 rows from work array, store into output array. */

  wsptr = workspace;
  for (ctr = 0; ctr < 2; ctr++) {
    outptr = output_buf[ctr] + output_col;
    /* It's not clear whether a zero row test is worthwhile here ... */

#ifndef NO_ZERO_ROW_TEST
    if (wsptr[1] == 0 && wsptr[3] == 0 && wsptr[5] == 0 && wsptr[7] == 0) {
      /* AC terms all zero */
      JSAMPLE dcval = range_limit[(int) DESCALE((INT32) wsptr[0], PASS1_BITS+3)
				  & RANGE_MASK];
      
      outptr[0] = dcval;
      outptr[1] = dcval;
      
      wsptr += DCTSIZE;		/* advance pointer to next row */
      continue;
    }
#endif
    
    /* Even part */
    
    tmp10 = ((INT32) wsptr[0]) << (CONST_BITS+2);
    
    /* Odd part */

    tmp0 = MULTIPLY((INT32) wsptr[7], - FIX_0_720959822) /* sqrt(2) * (c7-c5+c3-c1) */
	 + MULTIPLY((INT32) wsptr[5], FIX_0_850430095) /* sqrt(2) * (-c1+c3+c5+c7) */
	 + MULTIPLY((INT32) wsptr[3], - FIX_1_272758580) /* sqrt(2) * (-c1+c3-c5-c7) */
	 + MULTIPLY((INT32) wsptr[1], FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */

    /* Final output stage */
    
    outptr[0] = range_limit[(int) DESCALE(tmp10 + tmp0,
					  CONST_BITS+PASS1_BITS+3+2)
			    & RANGE_MASK];
    outptr[1] = range_limit[(int) DESCALE(tmp10 - tmp0,
					  CONST_BITS+PASS1_BITS+3+2)
			    & RANGE_MASK];
    
    wsptr += DCTSIZE;		/* advance pointer to next row */
  }
}
Пример #6
0
void j_rev_dct(DCTBLOCK data)
{
  int32 tmp0, tmp1, tmp2, tmp3;
  int32 tmp10, tmp11, tmp12, tmp13;
  int32 z1, z2, z3, z4, z5;
  register DCTELEM *dataptr;
  int rowctr;

  /* Pass 1: process rows. */
  /* Note results are scaled up by sqrt(8) compared to a true IDCT; */
  /* furthermore, we scale the results by 2**PASS1_BITS. */

  dataptr = data;
  for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) 
  {
    /* Even part: reverse the even part of the forward DCT. */
    /* The rotator is sqrt(2)*c(-6). */

    z2 = (int32) dataptr[2];
    z3 = (int32) dataptr[6];

    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);

    tmp0 = ((uint32) dataptr[0] + (uint32) dataptr[4]) << CONST_BITS;
    tmp1 = ((uint32) dataptr[0] - (uint32) dataptr[4]) << CONST_BITS;

    tmp10 = tmp0 + tmp3;
    tmp13 = tmp0 - tmp3;
    tmp11 = tmp1 + tmp2;
    tmp12 = tmp1 - tmp2;
    
    /* Odd part per figure 8; the matrix is unitary and hence its
     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
     */

    tmp0 = (int32) dataptr[7];
    tmp1 = (int32) dataptr[5];
    tmp2 = (int32) dataptr[3];
    tmp3 = (int32) dataptr[1];

    z1 = tmp0 + tmp3;
    z2 = tmp1 + tmp2;
    z3 = tmp0 + tmp2;
    z4 = tmp1 + tmp3;
    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
    
    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
    
    z3 += z5;
    z4 += z5;
    
    tmp0 += z1 + z3;
    tmp1 += z2 + z4;
    tmp2 += z2 + z3;
    tmp3 += z1 + z4;

    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */

    dataptr[0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
    dataptr[7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
    dataptr[1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
    dataptr[6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
    dataptr[2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
    dataptr[5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
    dataptr[3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
    dataptr[4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);

    dataptr += DCTSIZE;		/* advance pointer to next row */
  }

  /* Pass 2: process columns. */
  /* Note that we must descale the results by a factor of 8 == 2**3, */
  /* and also undo the PASS1_BITS scaling. */

  dataptr = data;
  for (rowctr = DCTSIZE-1; rowctr >= 0; rowctr--) {
    /* Even part: reverse the even part of the forward DCT. */
    /* The rotator is sqrt(2)*c(-6). */

    z2 = (int32) dataptr[DCTSIZE*2];
    z3 = (int32) dataptr[DCTSIZE*6];

    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);

    tmp0 = ((uint32) dataptr[DCTSIZE*0] + (uint32) dataptr[DCTSIZE*4]) << CONST_BITS;
    tmp1 = ((uint32) dataptr[DCTSIZE*0] - (uint32) dataptr[DCTSIZE*4]) << CONST_BITS;

    tmp10 = tmp0 + tmp3;
    tmp13 = tmp0 - tmp3;
    tmp11 = tmp1 + tmp2;
    tmp12 = tmp1 - tmp2;
    
    /* Odd part per figure 8; the matrix is unitary and hence its
     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
     */

    tmp0 = (int32) dataptr[DCTSIZE*7];
    tmp1 = (int32) dataptr[DCTSIZE*5];
    tmp2 = (int32) dataptr[DCTSIZE*3];
    tmp3 = (int32) dataptr[DCTSIZE*1];

    z1 = tmp0 + tmp3;
    z2 = tmp1 + tmp2;
    z3 = tmp0 + tmp2;
    z4 = tmp1 + tmp3;
    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
    
    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
    
    z3 += z5;
    z4 += z5;
    
    tmp0 += z1 + z3;
    tmp1 += z2 + z4;
    tmp2 += z2 + z3;
    tmp3 += z1 + z4;

    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */

    dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS+PASS1_BITS+1);
    dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS+PASS1_BITS+1);
    
    dataptr++;			/* advance pointer to next column */
  }
}
Пример #7
0
jpeg_idct_ifast (j_decompress_ptr cinfo, jpeg_component_info * compptr,
                 JCOEFPTR coef_block,
                 JSAMPARRAY output_buf, JDIMENSION output_col)
{
  DCTELEM tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
  DCTELEM tmp10, tmp11, tmp12, tmp13;
  DCTELEM z5, z10, z11, z12, z13;
  JCOEFPTR inptr;
  IFAST_MULT_TYPE * quantptr;
  int * wsptr;
  JSAMPROW outptr;
  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
  int ctr;
  int workspace[DCTSIZE2];      /* buffers data between passes */
  SHIFT_TEMPS                   /* for DESCALE */
  ISHIFT_TEMPS                  /* for IDESCALE */

  /* Pass 1: process columns from input, store into work array. */

  inptr = coef_block;
  quantptr = (IFAST_MULT_TYPE *) compptr->dct_table;
  wsptr = workspace;
  for (ctr = DCTSIZE; ctr > 0; ctr--) {
    /* Due to quantization, we will usually find that many of the input
     * coefficients are zero, especially the AC terms.  We can exploit this
     * by short-circuiting the IDCT calculation for any column in which all
     * the AC terms are zero.  In that case each output is equal to the
     * DC coefficient (with scale factor as needed).
     * With typical images and quantization tables, half or more of the
     * column DCT calculations can be simplified this way.
     */

    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
        inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*4] == 0 &&
        inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*6] == 0 &&
        inptr[DCTSIZE*7] == 0) {
      /* AC terms all zero */
      int dcval = (int) DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);

      wsptr[DCTSIZE*0] = dcval;
      wsptr[DCTSIZE*1] = dcval;
      wsptr[DCTSIZE*2] = dcval;
      wsptr[DCTSIZE*3] = dcval;
      wsptr[DCTSIZE*4] = dcval;
      wsptr[DCTSIZE*5] = dcval;
      wsptr[DCTSIZE*6] = dcval;
      wsptr[DCTSIZE*7] = dcval;

      inptr++;                  /* advance pointers to next column */
      quantptr++;
      wsptr++;
      continue;
    }

    /* Even part */

    tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
    tmp1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
    tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
    tmp3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);

    tmp10 = tmp0 + tmp2;        /* phase 3 */
    tmp11 = tmp0 - tmp2;

    tmp13 = tmp1 + tmp3;        /* phases 5-3 */
    tmp12 = MULTIPLY(tmp1 - tmp3, FIX_1_414213562) - tmp13; /* 2*c4 */

    tmp0 = tmp10 + tmp13;       /* phase 2 */
    tmp3 = tmp10 - tmp13;
    tmp1 = tmp11 + tmp12;
    tmp2 = tmp11 - tmp12;

    /* Odd part */

    tmp4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
    tmp5 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
    tmp6 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
    tmp7 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);

    z13 = tmp6 + tmp5;          /* phase 6 */
    z10 = tmp6 - tmp5;
    z11 = tmp4 + tmp7;
    z12 = tmp4 - tmp7;

    tmp7 = z11 + z13;           /* phase 5 */
    tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */

    z5 = MULTIPLY(z10 + z12, FIX_1_847759065); /* 2*c2 */
    tmp10 = MULTIPLY(z12, FIX_1_082392200) - z5; /* 2*(c2-c6) */
    tmp12 = MULTIPLY(z10, - FIX_2_613125930) + z5; /* -2*(c2+c6) */

    tmp6 = tmp12 - tmp7;        /* phase 2 */
    tmp5 = tmp11 - tmp6;
    tmp4 = tmp10 + tmp5;

    wsptr[DCTSIZE*0] = (int) (tmp0 + tmp7);
    wsptr[DCTSIZE*7] = (int) (tmp0 - tmp7);
    wsptr[DCTSIZE*1] = (int) (tmp1 + tmp6);
    wsptr[DCTSIZE*6] = (int) (tmp1 - tmp6);
    wsptr[DCTSIZE*2] = (int) (tmp2 + tmp5);
    wsptr[DCTSIZE*5] = (int) (tmp2 - tmp5);
    wsptr[DCTSIZE*4] = (int) (tmp3 + tmp4);
    wsptr[DCTSIZE*3] = (int) (tmp3 - tmp4);

    inptr++;                    /* advance pointers to next column */
    quantptr++;
    wsptr++;
  }

  /* Pass 2: process rows from work array, store into output array. */
  /* Note that we must descale the results by a factor of 8 == 2**3, */
  /* and also undo the PASS1_BITS scaling. */

  wsptr = workspace;
  for (ctr = 0; ctr < DCTSIZE; ctr++) {
    outptr = output_buf[ctr] + output_col;
    /* Rows of zeroes can be exploited in the same way as we did with columns.
     * However, the column calculation has created many nonzero AC terms, so
     * the simplification applies less often (typically 5% to 10% of the time).
     * On machines with very fast multiplication, it's possible that the
     * test takes more time than it's worth.  In that case this section
     * may be commented out.
     */

#ifndef NO_ZERO_ROW_TEST
    if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 && wsptr[4] == 0 &&
        wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
      /* AC terms all zero */
      JSAMPLE dcval = range_limit[IDESCALE(wsptr[0], PASS1_BITS+3)
                                  & RANGE_MASK];

      outptr[0] = dcval;
      outptr[1] = dcval;
      outptr[2] = dcval;
      outptr[3] = dcval;
      outptr[4] = dcval;
      outptr[5] = dcval;
      outptr[6] = dcval;
      outptr[7] = dcval;

      wsptr += DCTSIZE;         /* advance pointer to next row */
      continue;
    }
#endif

    /* Even part */

    tmp10 = ((DCTELEM) wsptr[0] + (DCTELEM) wsptr[4]);
    tmp11 = ((DCTELEM) wsptr[0] - (DCTELEM) wsptr[4]);

    tmp13 = ((DCTELEM) wsptr[2] + (DCTELEM) wsptr[6]);
    tmp12 = MULTIPLY((DCTELEM) wsptr[2] - (DCTELEM) wsptr[6], FIX_1_414213562)
            - tmp13;

    tmp0 = tmp10 + tmp13;
    tmp3 = tmp10 - tmp13;
    tmp1 = tmp11 + tmp12;
    tmp2 = tmp11 - tmp12;

    /* Odd part */

    z13 = (DCTELEM) wsptr[5] + (DCTELEM) wsptr[3];
    z10 = (DCTELEM) wsptr[5] - (DCTELEM) wsptr[3];
    z11 = (DCTELEM) wsptr[1] + (DCTELEM) wsptr[7];
    z12 = (DCTELEM) wsptr[1] - (DCTELEM) wsptr[7];

    tmp7 = z11 + z13;           /* phase 5 */
    tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */

    z5 = MULTIPLY(z10 + z12, FIX_1_847759065); /* 2*c2 */
    tmp10 = MULTIPLY(z12, FIX_1_082392200) - z5; /* 2*(c2-c6) */
    tmp12 = MULTIPLY(z10, - FIX_2_613125930) + z5; /* -2*(c2+c6) */

    tmp6 = tmp12 - tmp7;        /* phase 2 */
    tmp5 = tmp11 - tmp6;
    tmp4 = tmp10 + tmp5;

    /* Final output stage: scale down by a factor of 8 and range-limit */

    outptr[0] = range_limit[IDESCALE(tmp0 + tmp7, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[7] = range_limit[IDESCALE(tmp0 - tmp7, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[1] = range_limit[IDESCALE(tmp1 + tmp6, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[6] = range_limit[IDESCALE(tmp1 - tmp6, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[2] = range_limit[IDESCALE(tmp2 + tmp5, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[5] = range_limit[IDESCALE(tmp2 - tmp5, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[4] = range_limit[IDESCALE(tmp3 + tmp4, PASS1_BITS+3)
                            & RANGE_MASK];
    outptr[3] = range_limit[IDESCALE(tmp3 - tmp4, PASS1_BITS+3)
                            & RANGE_MASK];

    wsptr += DCTSIZE;           /* advance pointer to next row */
  }
}
void JPEG_SWIDCT_2X2(int16 *coef_block, uint8 *output_buf, const int32 *quantptr)
{
	int32 tmp0, tmp10, z1;
	int16 *inptr;
	int32 *wsptr;
	uint8 *outptr;
    int32 ctr;
	int32 workspace[DCTSIZE*2];	/* buffers data between passes */

	 /* Pass 1: process columns from input, store into work array. */

	inptr = coef_block;
	wsptr = workspace;
	for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) 
	{
		/* Don't bother to process columns 2,4,6 */
		if (ctr == DCTSIZE-2 || ctr == DCTSIZE-4 || ctr == DCTSIZE-6)
			continue;
		if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*3] == 0 &&
			inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*7] == 0) 
		{
			/* AC terms all zero; we need not examine terms 2,4,6 for 2x2 output */
			int32 dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
      
			wsptr[DCTSIZE*0] = dcval;
			wsptr[DCTSIZE*1] = dcval;
      
			continue;
		}
    
		/* Even part */
		z1 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
		tmp10 = z1 << (CONST_BITS+2);
    
		/* Odd part */
		z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
		tmp0 = MULTIPLY(z1, - FIX_0_720959822); /* sqrt(2) * (c7-c5+c3-c1) */
		z1 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
		tmp0 += MULTIPLY(z1, FIX_0_850430095); /* sqrt(2) * (-c1+c3+c5+c7) */
		z1 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
		tmp0 += MULTIPLY(z1, - FIX_1_272758580); /* sqrt(2) * (-c1+c3-c5-c7) */
		z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
		tmp0 += MULTIPLY(z1, FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */

		/* Final output stage */
		wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp0, CONST_BITS-PASS1_BITS+2);
		wsptr[DCTSIZE*1] = (int) DESCALE(tmp10 - tmp0, CONST_BITS-PASS1_BITS+2);
	}
  
	/* Pass 2: process 2 rows from work array, store into output array. */
	wsptr = workspace;
	for (ctr = 0; ctr < 2; ctr++) 
	{
		outptr = output_buf + ctr * 2;
		/* It's not clear whether a zero row test is worthwhile here ... */
		if (wsptr[1] == 0 && wsptr[3] == 0 && wsptr[5] == 0 && wsptr[7] == 0) 
		{
		  /* AC terms all zero */
		  uint8 dcval = s_pClip_table[(int32) DESCALE((int32) wsptr[0], PASS1_BITS+3)+128];
      
		  outptr[0] = dcval;
		  outptr[1] = dcval;
      
		  wsptr += DCTSIZE;		/* advance pointer to next row */
		  continue;
		}
    
		/* Even part */
        tmp10 = ((int32) wsptr[0]) << (CONST_BITS+2);
    
		/* Odd part */
	    tmp0 = MULTIPLY((int32) wsptr[7], - FIX_0_720959822) /* sqrt(2) * (c7-c5+c3-c1) */
			+ MULTIPLY((int32) wsptr[5], FIX_0_850430095) /* sqrt(2) * (-c1+c3+c5+c7) */
			+ MULTIPLY((int32) wsptr[3], - FIX_1_272758580) /* sqrt(2) * (-c1+c3-c5-c7) */
			+ MULTIPLY((int32) wsptr[1], FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */

		/* Final output stage */ 
		outptr[0] = s_pClip_table[(int32) DESCALE(tmp10 + tmp0, CONST_BITS+PASS1_BITS+3+2)+128];
		outptr[1] = s_pClip_table[(int32) DESCALE(tmp10 - tmp0, CONST_BITS+PASS1_BITS+3+2)+128];
    
		wsptr += DCTSIZE;		/* advance pointer to next row */
	}
}
void JPEG_SWIDCT_4X4(int16 *coef_block, uint8 *output_buf, const int32 *quantptr)
{
	int32 tmp0, tmp2, tmp10, tmp12;
	int32 z1, z2, z3, z4;
	int16 *inptr;
	int32 * wsptr;
	uint8 *outptr;
	int32 ctr;
	int32 workspace[DCTSIZE*4];	/* buffers data between passes */

	/* Pass 1: process columns from input, store into work array. */

	inptr = coef_block;
	wsptr = workspace;
	for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) 
	{
		/* Don't bother to process column 4, because second pass won't use it */
		if (ctr == DCTSIZE-4)
		  continue;
		if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
			inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*5] == 0 &&
			inptr[DCTSIZE*6] == 0 && inptr[DCTSIZE*7] == 0) 
		{
			/* AC terms all zero; we need not examine term 4 for 4x4 output */
			int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
      
			wsptr[DCTSIZE*0] = dcval;
			wsptr[DCTSIZE*1] = dcval;
			wsptr[DCTSIZE*2] = dcval;
			wsptr[DCTSIZE*3] = dcval;
      
			continue;
		}

		/* Even part */
    
		tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
		tmp0 <<= (CONST_BITS+1);
    
		z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
		z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);

		tmp2 = MULTIPLY(z2, FIX_1_847759065) + MULTIPLY(z3, -FIX_0_765366865);
    
		tmp10 = tmp0 + tmp2;
		tmp12 = tmp0 - tmp2;
    
		/* Odd part */
    
		z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
		z2 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
		z3 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
		z4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
    
		tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
			+ MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
			+ MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
			+ MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
    
		tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
			+ MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
			+ MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
			+ MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */

		/* Final output stage */
    
		wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp2, CONST_BITS-PASS1_BITS+1);
		wsptr[DCTSIZE*3] = (int) DESCALE(tmp10 - tmp2, CONST_BITS-PASS1_BITS+1);
		wsptr[DCTSIZE*1] = (int) DESCALE(tmp12 + tmp0, CONST_BITS-PASS1_BITS+1);
		wsptr[DCTSIZE*2] = (int) DESCALE(tmp12 - tmp0, CONST_BITS-PASS1_BITS+1);
	}

	/* Pass 2: process 4 rows from work array, store into output array. */

	wsptr = workspace;
	for (ctr = 0; ctr < 4; ctr++) 
	{
		outptr = output_buf + ctr * 4;
		/* It's not clear whether a zero row test is worthwhile here ... */

		if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 &&
			wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) 
		{
			/* AC terms all zero */
			uint8 dcval = s_pClip_table[(int32)DESCALE((int32) wsptr[0], PASS1_BITS+3)+128];
      
			outptr[0] = dcval;
			outptr[1] = dcval;
			outptr[2] = dcval;
			outptr[3] = dcval;
      
			wsptr += DCTSIZE;		/* advance pointer to next row */
			continue;
		}	
    
		/* Even part */
    
		tmp0 = ((int32) wsptr[0]) << (CONST_BITS+1);
    
		tmp2 = MULTIPLY((int32) wsptr[2], FIX_1_847759065)
			+ MULTIPLY((int32) wsptr[6], - FIX_0_765366865);
    
		tmp10 = tmp0 + tmp2;
		tmp12 = tmp0 - tmp2;
    
		/* Odd part */
    
		z1 = (int32) wsptr[7];
		z2 = (int32) wsptr[5];
		z3 = (int32) wsptr[3];
		z4 = (int32) wsptr[1];
    
		tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
			+ MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
			+ MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
			+ MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
    
		tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
			+ MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
			+ MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
			+ MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */

		/* Final output stage */
    
		outptr[0] = s_pClip_table[(int32) DESCALE(tmp10 + tmp2, CONST_BITS+PASS1_BITS+3+1)+128];
		outptr[3] = s_pClip_table[(int32) DESCALE(tmp10 - tmp2, CONST_BITS+PASS1_BITS+3+1)+128];
		outptr[1] = s_pClip_table[(int32) DESCALE(tmp12 + tmp0, CONST_BITS+PASS1_BITS+3+1)+128];
		outptr[2] = s_pClip_table[(int32) DESCALE(tmp12 - tmp0, CONST_BITS+PASS1_BITS+3+1)+128];
    
		wsptr += DCTSIZE;		/* advance pointer to next row */
	}
}