comparison i386/dsputil_mmx_avg.h @ 2967:ef2149182f1c libavcodec

COSMETICS: Remove all trailing whitespace.

2966:564788471dd4

 *
 * MMX optimization by Nick Kurshev <[e-mail]>
 * mostly rewritten by Michael Niedermayer <[e-mail]>
 * and improved by Zdenek Kabelac <[e-mail]>
 */
 
/* XXX: we use explicit registers to avoid a gcc 2.95.2 register asm
   clobber bug - now it will work with 2.95.2 and also with -fPIC
 */
static void DEF(put_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
}


static void DEF(put_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
}


static void DEF(avg_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory"); 
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}
 
/* GL: this function does incorrect rounding if overflow */
static void DEF(put_no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
    MOVQ_BONE(mm6);
    __asm __volatile(

	:"+g"(h), "+S"(pixels), "+D"(block)
	:"r" ((long)line_size)
	:"%"REG_a, "memory");
}

// Note this is not correctly rounded, but this function is only used for b frames so it doesnt matter 
static void DEF(avg_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
    MOVQ_BONE(mm6);
    __asm __volatile(
	"lea (%3, %3), %%"REG_a"	\n\t"

2967:ef2149182f1c

 *
 * MMX optimization by Nick Kurshev <[e-mail]>
 * mostly rewritten by Michael Niedermayer <[e-mail]>
 * and improved by Zdenek Kabelac <[e-mail]>
 */

/* XXX: we use explicit registers to avoid a gcc 2.95.2 register asm
   clobber bug - now it will work with 2.95.2 and also with -fPIC
 */
static void DEF(put_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
}


static void DEF(put_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
}


static void DEF(avg_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int dstStride, int src1Stride, int h)
{

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

	:"+m"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#else
	:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
	:"S"((long)src1Stride), "D"((long)dstStride)
	:"memory");
//the following should be used, though better not with gcc ...
/*	:"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
	:"r"(src1Stride), "r"(dstStride)
	:"memory");*/
}

/* GL: this function does incorrect rounding if overflow */
static void DEF(put_no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
    MOVQ_BONE(mm6);
    __asm __volatile(

	:"+g"(h), "+S"(pixels), "+D"(block)
	:"r" ((long)line_size)
	:"%"REG_a, "memory");
}

// Note this is not correctly rounded, but this function is only used for b frames so it doesnt matter
static void DEF(avg_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
    MOVQ_BONE(mm6);
    __asm __volatile(
	"lea (%3, %3), %%"REG_a"	\n\t"