comparison libpostproc/postprocess_template.c @ 2978:403183bbb505 libavcodec

Add some comments to #ifdef #else #endif blocks and fix wrong ones.

2977:49c6f104a76f

            else     return 1;
        }else{
            return 2;
        }
}
#endif

/**
 * Do a vertical low pass filter on the 8x16 block (only write to the 8x8 block in the middle)
 * using the 9-Tap Filter (1,1,2,2,4,2,2,1,1)/16
 */

		:
		: "r" (src), "r" ((long)stride), "m" (c->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else
	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

		src[l7]= (sums[6] + sums[8] + 2*src[l7])>>4;
		src[l8]= (sums[7] + sums[9] + 2*src[l8])>>4;

		src++;
	}
#endif
}
#endif //HAVE_ALTIVEC

#if 0
/**

		:
		: "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_c
	);
#else
 	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

			src[x+l6] -=v>>3;

		}
	}

#endif
}
#endif

/**
 * Experimental Filter 1
 * will not damage linear gradients
 * Flat blocks should look like they where passed through the (1,1,2,2,4,2,2,1,1) 9-Tap filter

		:
		: "r" (src), "r" ((long)stride), "m" (co->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else

 	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;

			src[l7] -=v>>3;

		}
		src++;
	}
#endif
}

#ifndef HAVE_ALTIVEC
static inline void RENAME(doVertDefFilter)(uint8_t src[], int stride, PPContext *c)
{

		"paddb %%mm3, %%mm2				\n\t"
		"pxor %%mm6, %%mm0				\n\t"
		"pxor %%mm6, %%mm2				\n\t"
		"movq %%mm0, (%%"REG_a", %1, 2)			\n\t"
		"movq %%mm2, (%0, %1, 4)			\n\t"
#endif

		"lea (%0, %1), %%"REG_a"			\n\t"
		"pcmpeqb %%mm6, %%mm6				\n\t" // -1
//	0	1	2	3	4	5	6	7
//	%0	%0+%1	%0+2%1	eax+2%1	%0+4%1	eax+4%1	ecx+%1	ecx+2%1

		: "+r" (src)
		: "r" ((long)stride), "m" (c->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else
	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

        		src[l4]-= d;
	        	src[l5]+= d;
		}
		src++;
	}
#endif
}
#endif //HAVE_ALTIVEC

#ifndef HAVE_ALTIVEC
static inline void RENAME(dering)(uint8_t src[], int stride, PPContext *c)

		"1:			\n\t"
		: : "r" (src), "r" ((long)stride), "m" (c->pQPb), "m"(c->pQPb2)
		: "%"REG_a, "%"REG_d, "%"REG_c
	);
#else
	int y;
	int min=255;
	int max=0;
	int avg;
	uint8_t *p;

			}
		}
//		src[0] = src[7]=src[stride*7]=src[stride*7 + 7]=255;
	}
#endif
#endif
}
#endif //HAVE_ALTIVEC

/**
 * Deinterlaces the given block by linearly interpolating every second line.

DEINT_CUBIC((%%REGd, %1), (%0, %1, 8), (%%REGd, %1, 4), (%%REGc), (%%REGc, %1, 2))

		: : "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_d, "%"REG_c
	);
#else
	int x;
	src+= stride*3;
	for(x=0; x<8; x++)
	{
		src[stride*3] = CLIP((-src[0]        + 9*src[stride*2] + 9*src[stride*4] - src[stride*6])>>4);
		src[stride*5] = CLIP((-src[stride*2] + 9*src[stride*4] + 9*src[stride*6] - src[stride*8])>>4);
		src[stride*7] = CLIP((-src[stride*4] + 9*src[stride*6] + 9*src[stride*8] - src[stride*10])>>4);
		src[stride*9] = CLIP((-src[stride*6] + 9*src[stride*8] + 9*src[stride*10] - src[stride*12])>>4);
		src++;
	}
#endif
}

/**
 * Deinterlaces the given block by filtering every second line with a (-1 4 2 4 -1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm0, (%2)				\n\t"
		: : "r" (src), "r" ((long)stride), "r"(tmp)
		: "%"REG_a, "%"REG_d
	);
#else
	int x;
	src+= stride*4;
	for(x=0; x<8; x++)
	{
		int t1= tmp[x];

		src[stride*7]= CLIP((-t2 + 4*src[stride*6] + 2*t1 + 4*src[stride*8] - src[stride*9] + 4)>>3);
		tmp[x]= t1;

		src++;
	}
#endif
}

/**
 * Deinterlaces the given block by filtering every line with a (-1 2 6 2 -1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm0, (%2)				\n\t"
		"movq %%mm1, (%3)				\n\t"
		: : "r" (src), "r" ((long)stride), "r"(tmp), "r"(tmp2)
		: "%"REG_a, "%"REG_d
	);
#else
	int x;
	src+= stride*4;
	for(x=0; x<8; x++)
	{
		int t1= tmp[x];

		tmp[x]= t3;
		tmp2[x]= t1;

		src++;
	}
#endif
}

/**
 * Deinterlaces the given block by filtering all lines with a (1 2 1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm1, (%2)				\n\t"

		: : "r" (src), "r" ((long)stride), "r" (tmp)
		: "%"REG_a, "%"REG_d
	);
#else
	int a, b, c, x;
	src+= 4*stride;

	for(x=0; x<2; x++){
		a= *(uint32_t*)&tmp[stride*0];

		*(uint32_t*)&tmp[stride*0]= c;
		src += 4;
		tmp += 4;
	}
#endif
}

/**
 * Deinterlaces the given block by applying a median filter to every second line.
 * will be called for every 8x8 block and can read & write from line 4-15,

MEDIAN((%%REGd, %1), (%%REGd, %1, 2), (%0, %1, 8))

		: : "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_d
	);
#endif // MMX
#else
	int x, y;
	src+= 4*stride;
	// FIXME - there should be a way to do a few columns in parallel like w/mmx
	for(x=0; x<8; x++)
	{

			colsrc[stride  ] = (a|(d^f)) & (b|(d^e)) & (c|(e^f));
			colsrc += stride*2;
		}
		src++;
	}
#endif
}

#ifdef HAVE_MMX
/**
 * transposes and shift the given 8x8 Block into dst1 and dst2

	:: "r" (dst), "r" ((long)dstStride), "r" (src)
	: "%"REG_a, "%"REG_d
	);
}
#endif
//static long test=0;

#ifndef HAVE_ALTIVEC
static inline void RENAME(tempNoiseReducer)(uint8_t *src, int stride,
				    uint8_t *tempBlured, uint32_t *tempBluredPast, int *maxNoise)

		"paddw %%mm4, %%mm0				\n\t"
		"psadbw (%1, %%"REG_c"), %%mm7			\n\t" // |L7-R7|
		"paddw %%mm5, %%mm6				\n\t"
		"paddw %%mm7, %%mm6				\n\t"
		"paddw %%mm6, %%mm0				\n\t"
#else
#if defined (FAST_L2_DIFF)
		"pcmpeqb %%mm7, %%mm7				\n\t"
		"movq "MANGLE(b80)", %%mm6			\n\t"
		"pxor %%mm0, %%mm0				\n\t"
#define REAL_L2_DIFF_CORE(a, b)\

		"pmaddwd %%mm2, %%mm2				\n\t"\
		"paddd %%mm2, %%mm5				\n\t"\
		"psrld $14, %%mm5				\n\t"\
		"paddd %%mm5, %%mm0				\n\t"

#else
		"pxor %%mm7, %%mm7				\n\t"
		"pxor %%mm0, %%mm0				\n\t"
#define REAL_L2_DIFF_CORE(a, b)\
		"movq " #a ", %%mm5				\n\t"\
		"movq " #b ", %%mm2				\n\t"\

		"pmaddwd %%mm5, %%mm5				\n\t"\
		"pmaddwd %%mm1, %%mm1				\n\t"\
		"paddd %%mm1, %%mm5				\n\t"\
		"paddd %%mm5, %%mm0				\n\t"

#endif

#define L2_DIFF_CORE(a, b)  REAL_L2_DIFF_CORE(a, b)

L2_DIFF_CORE((%0), (%1))
L2_DIFF_CORE((%0, %2), (%1, %2))

L2_DIFF_CORE((%0, %2, 4), (%1, %2, 4))
L2_DIFF_CORE((%0, %%REGd), (%1, %%REGd))
L2_DIFF_CORE((%0, %%REGa,2), (%1, %%REGa,2))
L2_DIFF_CORE((%0, %%REGc), (%1, %%REGc))

#endif

		"movq %%mm0, %%mm4				\n\t"
		"psrlq $32, %%mm0				\n\t"
		"paddd %%mm0, %%mm4				\n\t"
		"movd %%mm4, %%ecx				\n\t"

		:: "r" (src), "r" (tempBlured), "r"((long)stride), "m" (tempBluredPast)
		: "%"REG_a, "%"REG_d, "%"REG_c, "memory"
		);
//printf("%d\n", test);
#else
{
	int y;
	int d=0;
//	int sysd=0;
	int i;

				}
			}
		}
	}
}
#endif
}
#endif //HAVE_ALTIVEC

#ifdef HAVE_MMX
/**

						"packuswb %%mm5, %%mm0	\n\t"\
						"packuswb %%mm6, %%mm1	\n\t"\
						"movq %%mm0, " #dst1 "	\n\t"\
						"movq %%mm1, " #dst2 "	\n\t"\

#endif //!HAVE_MMX2
#define SCALED_CPY(src1, src2, dst1, dst2)\
   REAL_SCALED_CPY(src1, src2, dst1, dst2)

SCALED_CPY((%2)       , (%2, %4)      , (%3)       , (%3, %5))
SCALED_CPY((%2, %4, 2), (%%REGa, %4, 2), (%3, %5, 2), (%%REGd, %5, 2))

						"r"(dst),
						"r" ((long)srcStride),
						"r" ((long)dstStride)
						: "%"REG_d
					);
#else
				for(i=0; i<8; i++)
					memcpy(	&(dst[dstStride*i]),
						&(src[srcStride*i]), BLOCK_SIZE);
#endif
	}
	else
	{
#ifdef HAVE_MMX
					asm volatile(

						"r" (dst),
						"r" ((long)srcStride),
						"r" ((long)dstStride)
						: "%"REG_a, "%"REG_d
					);
#else
				for(i=0; i<8; i++)
					memcpy(	&(dst[dstStride*i]),
						&(src[srcStride*i]), BLOCK_SIZE);
#endif
	}
}

/**
 * Duplicates the given 8 src pixels ? times upward

						RENAME(doHorizDefFilter)(dstBlock-4, stride, &c);
#endif
				}else if(mode & H_A_DEBLOCK){
					RENAME(do_a_deblock)(dstBlock-8, 1, stride, &c);
				}
#endif
				if(mode & DERING)
				{
				//FIXME filter first line
					if(y>0) RENAME(dering)(dstBlock - stride - 8, stride, &c);
				}

2978:403183bbb505

            else     return 1;
        }else{
            return 2;
        }
}
#endif //HAVE_MMX

/**
 * Do a vertical low pass filter on the 8x16 block (only write to the 8x8 block in the middle)
 * using the 9-Tap Filter (1,1,2,2,4,2,2,1,1)/16
 */

		:
		: "r" (src), "r" ((long)stride), "m" (c->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

		src[l7]= (sums[6] + sums[8] + 2*src[l7])>>4;
		src[l8]= (sums[7] + sums[9] + 2*src[l8])>>4;

		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}
#endif //HAVE_ALTIVEC

#if 0
/**

		:
		: "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
 	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

			src[x+l6] -=v>>3;

		}
	}

#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}
#endif //0

/**
 * Experimental Filter 1
 * will not damage linear gradients
 * Flat blocks should look like they where passed through the (1,1,2,2,4,2,2,1,1) 9-Tap filter

		:
		: "r" (src), "r" ((long)stride), "m" (co->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)

 	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;

			src[l7] -=v>>3;

		}
		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}

#ifndef HAVE_ALTIVEC
static inline void RENAME(doVertDefFilter)(uint8_t src[], int stride, PPContext *c)
{

		"paddb %%mm3, %%mm2				\n\t"
		"pxor %%mm6, %%mm0				\n\t"
		"pxor %%mm6, %%mm2				\n\t"
		"movq %%mm0, (%%"REG_a", %1, 2)			\n\t"
		"movq %%mm2, (%0, %1, 4)			\n\t"
#endif //0

		"lea (%0, %1), %%"REG_a"			\n\t"
		"pcmpeqb %%mm6, %%mm6				\n\t" // -1
//	0	1	2	3	4	5	6	7
//	%0	%0+%1	%0+2%1	eax+2%1	%0+4%1	eax+4%1	ecx+%1	ecx+2%1

		: "+r" (src)
		: "r" ((long)stride), "m" (c->pQPb)
		: "%"REG_a, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	const int l1= stride;
	const int l2= stride + l1;
	const int l3= stride + l2;
	const int l4= stride + l3;
	const int l5= stride + l4;

        		src[l4]-= d;
	        	src[l5]+= d;
		}
		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}
#endif //HAVE_ALTIVEC

#ifndef HAVE_ALTIVEC
static inline void RENAME(dering)(uint8_t src[], int stride, PPContext *c)

		"1:			\n\t"
		: : "r" (src), "r" ((long)stride), "m" (c->pQPb), "m"(c->pQPb2)
		: "%"REG_a, "%"REG_d, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	int y;
	int min=255;
	int max=0;
	int avg;
	uint8_t *p;

			}
		}
//		src[0] = src[7]=src[stride*7]=src[stride*7 + 7]=255;
	}
#endif
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}
#endif //HAVE_ALTIVEC

/**
 * Deinterlaces the given block by linearly interpolating every second line.

DEINT_CUBIC((%%REGd, %1), (%0, %1, 8), (%%REGd, %1, 4), (%%REGc), (%%REGc, %1, 2))

		: : "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_d, "%"REG_c
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	int x;
	src+= stride*3;
	for(x=0; x<8; x++)
	{
		src[stride*3] = CLIP((-src[0]        + 9*src[stride*2] + 9*src[stride*4] - src[stride*6])>>4);
		src[stride*5] = CLIP((-src[stride*2] + 9*src[stride*4] + 9*src[stride*6] - src[stride*8])>>4);
		src[stride*7] = CLIP((-src[stride*4] + 9*src[stride*6] + 9*src[stride*8] - src[stride*10])>>4);
		src[stride*9] = CLIP((-src[stride*6] + 9*src[stride*8] + 9*src[stride*10] - src[stride*12])>>4);
		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}

/**
 * Deinterlaces the given block by filtering every second line with a (-1 4 2 4 -1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm0, (%2)				\n\t"
		: : "r" (src), "r" ((long)stride), "r"(tmp)
		: "%"REG_a, "%"REG_d
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	int x;
	src+= stride*4;
	for(x=0; x<8; x++)
	{
		int t1= tmp[x];

		src[stride*7]= CLIP((-t2 + 4*src[stride*6] + 2*t1 + 4*src[stride*8] - src[stride*9] + 4)>>3);
		tmp[x]= t1;

		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}

/**
 * Deinterlaces the given block by filtering every line with a (-1 2 6 2 -1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm0, (%2)				\n\t"
		"movq %%mm1, (%3)				\n\t"
		: : "r" (src), "r" ((long)stride), "r"(tmp), "r"(tmp2)
		: "%"REG_a, "%"REG_d
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	int x;
	src+= stride*4;
	for(x=0; x<8; x++)
	{
		int t1= tmp[x];

		tmp[x]= t3;
		tmp2[x]= t1;

		src++;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}

/**
 * Deinterlaces the given block by filtering all lines with a (1 2 1) filter.
 * will be called for every 8x8 block and can read & write from line 4-15

		"movq %%mm1, (%2)				\n\t"

		: : "r" (src), "r" ((long)stride), "r" (tmp)
		: "%"REG_a, "%"REG_d
	);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
	int a, b, c, x;
	src+= 4*stride;

	for(x=0; x<2; x++){
		a= *(uint32_t*)&tmp[stride*0];

		*(uint32_t*)&tmp[stride*0]= c;
		src += 4;
		tmp += 4;
	}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}

/**
 * Deinterlaces the given block by applying a median filter to every second line.
 * will be called for every 8x8 block and can read & write from line 4-15,

MEDIAN((%%REGd, %1), (%%REGd, %1, 2), (%0, %1, 8))

		: : "r" (src), "r" ((long)stride)
		: "%"REG_a, "%"REG_d
	);
#endif //HAVE_MMX2
#else //HAVE_MMX
	int x, y;
	src+= 4*stride;
	// FIXME - there should be a way to do a few columns in parallel like w/mmx
	for(x=0; x<8; x++)
	{

			colsrc[stride  ] = (a|(d^f)) & (b|(d^e)) & (c|(e^f));
			colsrc += stride*2;
		}
		src++;
	}
#endif //HAVE_MMX
}

#ifdef HAVE_MMX
/**
 * transposes and shift the given 8x8 Block into dst1 and dst2

	:: "r" (dst), "r" ((long)dstStride), "r" (src)
	: "%"REG_a, "%"REG_d
	);
}
#endif //HAVE_MMX
//static long test=0;

#ifndef HAVE_ALTIVEC
static inline void RENAME(tempNoiseReducer)(uint8_t *src, int stride,
				    uint8_t *tempBlured, uint32_t *tempBluredPast, int *maxNoise)

		"paddw %%mm4, %%mm0				\n\t"
		"psadbw (%1, %%"REG_c"), %%mm7			\n\t" // |L7-R7|
		"paddw %%mm5, %%mm6				\n\t"
		"paddw %%mm7, %%mm6				\n\t"
		"paddw %%mm6, %%mm0				\n\t"
#else //L1_DIFF
#if defined (FAST_L2_DIFF)
		"pcmpeqb %%mm7, %%mm7				\n\t"
		"movq "MANGLE(b80)", %%mm6			\n\t"
		"pxor %%mm0, %%mm0				\n\t"
#define REAL_L2_DIFF_CORE(a, b)\

		"pmaddwd %%mm2, %%mm2				\n\t"\
		"paddd %%mm2, %%mm5				\n\t"\
		"psrld $14, %%mm5				\n\t"\
		"paddd %%mm5, %%mm0				\n\t"

#else //defined (FAST_L2_DIFF)
		"pxor %%mm7, %%mm7				\n\t"
		"pxor %%mm0, %%mm0				\n\t"
#define REAL_L2_DIFF_CORE(a, b)\
		"movq " #a ", %%mm5				\n\t"\
		"movq " #b ", %%mm2				\n\t"\

		"pmaddwd %%mm5, %%mm5				\n\t"\
		"pmaddwd %%mm1, %%mm1				\n\t"\
		"paddd %%mm1, %%mm5				\n\t"\
		"paddd %%mm5, %%mm0				\n\t"

#endif //defined (FAST_L2_DIFF)

#define L2_DIFF_CORE(a, b)  REAL_L2_DIFF_CORE(a, b)

L2_DIFF_CORE((%0), (%1))
L2_DIFF_CORE((%0, %2), (%1, %2))

L2_DIFF_CORE((%0, %2, 4), (%1, %2, 4))
L2_DIFF_CORE((%0, %%REGd), (%1, %%REGd))
L2_DIFF_CORE((%0, %%REGa,2), (%1, %%REGa,2))
L2_DIFF_CORE((%0, %%REGc), (%1, %%REGc))

#endif //L1_DIFF

		"movq %%mm0, %%mm4				\n\t"
		"psrlq $32, %%mm0				\n\t"
		"paddd %%mm0, %%mm4				\n\t"
		"movd %%mm4, %%ecx				\n\t"

		:: "r" (src), "r" (tempBlured), "r"((long)stride), "m" (tempBluredPast)
		: "%"REG_a, "%"REG_d, "%"REG_c, "memory"
		);
//printf("%d\n", test);
#else //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
{
	int y;
	int d=0;
//	int sysd=0;
	int i;

				}
			}
		}
	}
}
#endif //defined (HAVE_MMX2) || defined (HAVE_3DNOW)
}
#endif //HAVE_ALTIVEC

#ifdef HAVE_MMX
/**

						"packuswb %%mm5, %%mm0	\n\t"\
						"packuswb %%mm6, %%mm1	\n\t"\
						"movq %%mm0, " #dst1 "	\n\t"\
						"movq %%mm1, " #dst2 "	\n\t"\

#endif //HAVE_MMX2
#define SCALED_CPY(src1, src2, dst1, dst2)\
   REAL_SCALED_CPY(src1, src2, dst1, dst2)

SCALED_CPY((%2)       , (%2, %4)      , (%3)       , (%3, %5))
SCALED_CPY((%2, %4, 2), (%%REGa, %4, 2), (%3, %5, 2), (%%REGd, %5, 2))

						"r"(dst),
						"r" ((long)srcStride),
						"r" ((long)dstStride)
						: "%"REG_d
					);
#else //HAVE_MMX
				for(i=0; i<8; i++)
					memcpy(	&(dst[dstStride*i]),
						&(src[srcStride*i]), BLOCK_SIZE);
#endif //HAVE_MMX
	}
	else
	{
#ifdef HAVE_MMX
					asm volatile(

						"r" (dst),
						"r" ((long)srcStride),
						"r" ((long)dstStride)
						: "%"REG_a, "%"REG_d
					);
#else //HAVE_MMX
				for(i=0; i<8; i++)
					memcpy(	&(dst[dstStride*i]),
						&(src[srcStride*i]), BLOCK_SIZE);
#endif //HAVE_MMX
	}
}

/**
 * Duplicates the given 8 src pixels ? times upward

						RENAME(doHorizDefFilter)(dstBlock-4, stride, &c);
#endif
				}else if(mode & H_A_DEBLOCK){
					RENAME(do_a_deblock)(dstBlock-8, 1, stride, &c);
				}
#endif //HAVE_MMX
				if(mode & DERING)
				{
				//FIXME filter first line
					if(y>0) RENAME(dering)(dstBlock - stride - 8, stride, &c);
				}