📄 rgb2rgb_template.c
字号:
"movq (%1, %%"REG_a"), %%mm1 \n\t"
"movq 1(%0, %%"REG_a"), %%mm2 \n\t"
"movq 1(%1, %%"REG_a"), %%mm3 \n\t"
"movq -1(%0, %%"REG_a"), %%mm4 \n\t"
"movq -1(%1, %%"REG_a"), %%mm5 \n\t"
PAVGB" %%mm0, %%mm5 \n\t"
PAVGB" %%mm0, %%mm3 \n\t"
PAVGB" %%mm0, %%mm5 \n\t"
PAVGB" %%mm0, %%mm3 \n\t"
PAVGB" %%mm1, %%mm4 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
PAVGB" %%mm1, %%mm4 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
"movq %%mm5, %%mm7 \n\t"
"movq %%mm4, %%mm6 \n\t"
"punpcklbw %%mm3, %%mm5 \n\t"
"punpckhbw %%mm3, %%mm7 \n\t"
"punpcklbw %%mm2, %%mm4 \n\t"
"punpckhbw %%mm2, %%mm6 \n\t"
#if 1
MOVNTQ" %%mm5, (%2, %%"REG_a", 2)\n\t"
MOVNTQ" %%mm7, 8(%2, %%"REG_a", 2)\n\t"
MOVNTQ" %%mm4, (%3, %%"REG_a", 2)\n\t"
MOVNTQ" %%mm6, 8(%3, %%"REG_a", 2)\n\t"
#else
"movq %%mm5, (%2, %%"REG_a", 2) \n\t"
"movq %%mm7, 8(%2, %%"REG_a", 2)\n\t"
"movq %%mm4, (%3, %%"REG_a", 2) \n\t"
"movq %%mm6, 8(%3, %%"REG_a", 2)\n\t"
#endif
"add $8, %%"REG_a" \n\t"
" js 1b \n\t"
:: "r" (src + mmxSize ), "r" (src + srcStride + mmxSize ),
"r" (dst + mmxSize*2), "r" (dst + dstStride + mmxSize*2),
"g" (-mmxSize)
: "%"REG_a
);
#else
const long mmxSize=1;
#endif
dst[0 ]= (3*src[0] + src[srcStride])>>2;
dst[dstStride]= ( src[0] + 3*src[srcStride])>>2;
for(x=mmxSize-1; x<srcWidth-1; x++){
dst[2*x +1]= (3*src[x+0] + src[x+srcStride+1])>>2;
dst[2*x+dstStride+2]= ( src[x+0] + 3*src[x+srcStride+1])>>2;
dst[2*x+dstStride+1]= ( src[x+1] + 3*src[x+srcStride ])>>2;
dst[2*x +2]= (3*src[x+1] + src[x+srcStride ])>>2;
}
dst[srcWidth*2 -1 ]= (3*src[srcWidth-1] + src[srcWidth-1 + srcStride])>>2;
dst[srcWidth*2 -1 + dstStride]= ( src[srcWidth-1] + 3*src[srcWidth-1 + srcStride])>>2;
dst+=dstStride*2;
src+=srcStride;
}
// last line
#if 1
dst[0]= src[0];
for(x=0; x<srcWidth-1; x++){
dst[2*x+1]= (3*src[x] + src[x+1])>>2;
dst[2*x+2]= ( src[x] + 3*src[x+1])>>2;
}
dst[2*srcWidth-1]= src[srcWidth-1];
#else
for(x=0; x<srcWidth; x++){
dst[2*x+0]=
dst[2*x+1]= src[x];
}
#endif
#ifdef HAVE_MMX
asm volatile( EMMS" \n\t"
SFENCE" \n\t"
:::"memory");
#endif
}
/**
*
* height should be a multiple of 2 and width should be a multiple of 16 (if this is a
* problem for anyone then tell me, and ill fix it)
* chrominance data is only taken from every secound line others are ignored FIXME write HQ version
*/
static inline void RENAME(uyvytoyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
long width, long height,
stride_t lumStride, stride_t chromStride, stride_t srcStride)
{
long y;
const long chromWidth= width>>1;
for(y=0; y<height; y+=2)
{
#ifdef HAVE_MMX
asm volatile(
"xorl %%eax, %%eax \n\t"
"pcmpeqw %%mm7, %%mm7 \n\t"
"psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
".balign 16 \n\t"
"1: \n\t"
PREFETCH" 64(%0, %%eax, 4) \n\t"
"movq (%0, %%eax, 4), %%mm0 \n\t" // UYVY UYVY(0)
"movq 8(%0, %%eax, 4), %%mm1 \n\t" // UYVY UYVY(4)
"movq %%mm0, %%mm2 \n\t" // UYVY UYVY(0)
"movq %%mm1, %%mm3 \n\t" // UYVY UYVY(4)
"pand %%mm7, %%mm0 \n\t" // U0V0 U0V0(0)
"pand %%mm7, %%mm1 \n\t" // U0V0 U0V0(4)
"psrlw $8, %%mm2 \n\t" // Y0Y0 Y0Y0(0)
"psrlw $8, %%mm3 \n\t" // Y0Y0 Y0Y0(4)
"packuswb %%mm1, %%mm0 \n\t" // UVUV UVUV(0)
"packuswb %%mm3, %%mm2 \n\t" // YYYY YYYY(0)
MOVNTQ" %%mm2, (%1, %%eax, 2) \n\t"
"movq 16(%0, %%eax, 4), %%mm1 \n\t" // UYVY UYVY(8)
"movq 24(%0, %%eax, 4), %%mm2 \n\t" // UYVY UYVY(12)
"movq %%mm1, %%mm3 \n\t" // UYVY UYVY(8)
"movq %%mm2, %%mm4 \n\t" // UYVY UYVY(12)
"pand %%mm7, %%mm1 \n\t" // U0V0 U0V0(8)
"pand %%mm7, %%mm2 \n\t" // U0V0 U0V0(12)
"psrlw $8, %%mm3 \n\t" // Y0Y0 Y0Y0(8)
"psrlw $8, %%mm4 \n\t" // Y0Y0 Y0Y0(12)
"packuswb %%mm2, %%mm1 \n\t" // UVUV UVUV(8)
"packuswb %%mm4, %%mm3 \n\t" // YYYY YYYY(8)
MOVNTQ" %%mm3, 8(%1, %%eax, 2) \n\t"
"movq %%mm0, %%mm2 \n\t" // UVUV UVUV(0)
"movq %%mm1, %%mm3 \n\t" // UVUV UVUV(8)
"psrlw $8, %%mm0 \n\t" // V0V0 V0V0(0)
"psrlw $8, %%mm1 \n\t" // V0V0 V0V0(8)
"pand %%mm7, %%mm2 \n\t" // U0U0 U0U0(0)
"pand %%mm7, %%mm3 \n\t" // U0U0 U0U0(8)
"packuswb %%mm1, %%mm0 \n\t" // VVVV VVVV(0)
"packuswb %%mm3, %%mm2 \n\t" // UUUU UUUU(0)
MOVNTQ" %%mm0, (%3, %%eax) \n\t"
MOVNTQ" %%mm2, (%2, %%eax) \n\t"
"addl $8, %%eax \n\t"
"cmpl %4, %%eax \n\t"
" jb 1b \n\t"
::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
: "memory", "%eax"
);
ydst += lumStride;
src += srcStride;
asm volatile(
"xorl %%eax, %%eax \n\t"
".balign 16 \n\t"
"1: \n\t"
PREFETCH" 64(%0, %%eax, 4) \n\t"
"movq (%0, %%eax, 4), %%mm0 \n\t" // YUYV YUYV(0)
"movq 8(%0, %%eax, 4), %%mm1 \n\t" // YUYV YUYV(4)
"movq 16(%0, %%eax, 4), %%mm2 \n\t" // YUYV YUYV(8)
"movq 24(%0, %%eax, 4), %%mm3 \n\t" // YUYV YUYV(12)
"psrlw $8, %%mm0 \n\t" // Y0Y0 Y0Y0(0)
"psrlw $8, %%mm1 \n\t" // Y0Y0 Y0Y0(4)
"psrlw $8, %%mm2 \n\t" // Y0Y0 Y0Y0(8)
"psrlw $8, %%mm3 \n\t" // Y0Y0 Y0Y0(12)
"packuswb %%mm1, %%mm0 \n\t" // YYYY YYYY(0)
"packuswb %%mm3, %%mm2 \n\t" // YYYY YYYY(8)
MOVNTQ" %%mm0, (%1, %%eax, 2) \n\t"
MOVNTQ" %%mm2, 8(%1, %%eax, 2) \n\t"
"addl $8, %%eax \n\t"
"cmpl %4, %%eax \n\t"
" jb 1b \n\t"
::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
: "memory", "%eax"
);
#else
long i;
for(i=0; i<chromWidth; i++)
{
udst[i] = src[4*i+0];
ydst[2*i+0] = src[4*i+1];
vdst[i] = src[4*i+2];
ydst[2*i+1] = src[4*i+3];
}
ydst += lumStride;
src += srcStride;
for(i=0; i<chromWidth; i++)
{
ydst[2*i+0] = src[4*i+1];
ydst[2*i+1] = src[4*i+3];
}
#endif
udst += chromStride;
vdst += chromStride;
ydst += lumStride;
src += srcStride;
}
#ifdef HAVE_MMX
asm volatile( EMMS" \n\t"
SFENCE" \n\t"
:::"memory");
#endif
}
/**
*
* height should be a multiple of 2 and width should be a multiple of 2 (if this is a
* problem for anyone then tell me, and ill fix it)
* chrominance data is only taken from every secound line others are ignored in the C version FIXME write HQ version
*/
static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
long width, long height,
stride_t lumStride, stride_t chromStride, stride_t srcStride)
{
long y;
const long chromWidth= width>>1;
#ifdef HAVE_MMX
for(y=0; y<height-2; y+=2)
{
long i;
for(i=0; i<2; i++)
{
asm volatile(
"mov %2, %%"REG_a" \n\t"
"movq "MANGLE(bgr2YCoeff)", %%mm6 \n\t"
"movq "MANGLE(w1111)", %%mm5 \n\t"
"pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t"
".balign 16 \n\t"
"1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t"
"movd (%0, %%"REG_b"), %%mm0 \n\t"
"movd 3(%0, %%"REG_b"), %%mm1 \n\t"
"punpcklbw %%mm7, %%mm0 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"movd 6(%0, %%"REG_b"), %%mm2 \n\t"
"movd 9(%0, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"pmaddwd %%mm6, %%mm0 \n\t"
"pmaddwd %%mm6, %%mm1 \n\t"
"pmaddwd %%mm6, %%mm2 \n\t"
"pmaddwd %%mm6, %%mm3 \n\t"
#ifndef FAST_BGR2YV12
"psrad $8, %%mm0 \n\t"
"psrad $8, %%mm1 \n\t"
"psrad $8, %%mm2 \n\t"
"psrad $8, %%mm3 \n\t"
#endif
"packssdw %%mm1, %%mm0 \n\t"
"packssdw %%mm3, %%mm2 \n\t"
"pmaddwd %%mm5, %%mm0 \n\t"
"pmaddwd %%mm5, %%mm2 \n\t"
"packssdw %%mm2, %%mm0 \n\t"
"psraw $7, %%mm0 \n\t"
"movd 12(%0, %%"REG_b"), %%mm4 \n\t"
"movd 15(%0, %%"REG_b"), %%mm1 \n\t"
"punpcklbw %%mm7, %%mm4 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"movd 18(%0, %%"REG_b"), %%mm2 \n\t"
"movd 21(%0, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"pmaddwd %%mm6, %%mm4 \n\t"
"pmaddwd %%mm6, %%mm1 \n\t"
"pmaddwd %%mm6, %%mm2 \n\t"
"pmaddwd %%mm6, %%mm3 \n\t"
#ifndef FAST_BGR2YV12
"psrad $8, %%mm4 \n\t"
"psrad $8, %%mm1 \n\t"
"psrad $8, %%mm2 \n\t"
"psrad $8, %%mm3 \n\t"
#endif
"packssdw %%mm1, %%mm4 \n\t"
"packssdw %%mm3, %%mm2 \n\t"
"pmaddwd %%mm5, %%mm4 \n\t"
"pmaddwd %%mm5, %%mm2 \n\t"
"add $24, %%"REG_b" \n\t"
"packssdw %%mm2, %%mm4 \n\t"
"psraw $7, %%mm4 \n\t"
"packuswb %%mm4, %%mm0 \n\t"
"paddusb "MANGLE(bgr2YOffset)", %%mm0 \n\t"
MOVNTQ" %%mm0, (%1, %%"REG_a") \n\t"
"add $8, %%"REG_a" \n\t"
" js 1b \n\t"
: : "r" (src+width*3), "r" (ydst+width), "g" (-width)
: "%"REG_a, "%"REG_b
);
ydst += lumStride;
src += srcStride;
}
src -= srcStride*2;
asm volatile(
"mov %4, %%"REG_a" \n\t"
"movq "MANGLE(w1111)", %%mm5 \n\t"
"movq "MANGLE(bgr2UCoeff)", %%mm6 \n\t"
"pxor %%mm7, %%mm7 \n\t"
"lea (%%"REG_a", %%"REG_a", 2), %%"REG_b"\n\t"
"add %%"REG_b", %%"REG_b" \n\t"
".balign 16 \n\t"
"1: \n\t"
PREFETCH" 64(%0, %%"REG_b") \n\t"
PREFETCH" 64(%1, %%"REG_b") \n\t"
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
"movq (%0, %%"REG_b"), %%mm0 \n\t"
"movq (%1, %%"REG_b"), %%mm1 \n\t"
"movq 6(%0, %%"REG_b"), %%mm2 \n\t"
"movq 6(%1, %%"REG_b"), %%mm3 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm3, %%mm2 \n\t"
"movq %%mm0, %%mm1 \n\t"
"movq %%mm2, %%mm3 \n\t"
"psrlq $24, %%mm0 \n\t"
"psrlq $24, %%mm2 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm3, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm0 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
#else
"movd (%0, %%"REG_b"), %%mm0 \n\t"
"movd (%1, %%"REG_b"), %%mm1 \n\t"
"movd 3(%0, %%"REG_b"), %%mm2 \n\t"
"movd 3(%1, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm0 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"paddw %%mm1, %%mm0 \n\t"
"paddw %%mm3, %%mm2 \n\t"
"paddw %%mm2, %%mm0 \n\t"
"movd 6(%0, %%"REG_b"), %%mm4 \n\t"
"movd 6(%1, %%"REG_b"), %%mm1 \n\t"
"movd 9(%0, %%"REG_b"), %%mm2 \n\t"
"movd 9(%1, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm4 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"paddw %%mm1, %%mm4 \n\t"
"paddw %%mm3, %%mm2 \n\t"
"paddw %%mm4, %%mm2 \n\t"
"psrlw $2, %%mm0 \n\t"
"psrlw $2, %%mm2 \n\t"
#endif
"movq "MANGLE(bgr2VCoeff)", %%mm1 \n\t"
"movq "MANGLE(bgr2VCoeff)", %%mm3 \n\t"
"pmaddwd %%mm0, %%mm1 \n\t"
"pmaddwd %%mm2, %%mm3 \n\t"
"pmaddwd %%mm6, %%mm0 \n\t"
"pmaddwd %%mm6, %%mm2 \n\t"
#ifndef FAST_BGR2YV12
"psrad $8, %%mm0 \n\t"
"psrad $8, %%mm1 \n\t"
"psrad $8, %%mm2 \n\t"
"psrad $8, %%mm3 \n\t"
#endif
"packssdw %%mm2, %%mm0 \n\t"
"packssdw %%mm3, %%mm1 \n\t"
"pmaddwd %%mm5, %%mm0 \n\t"
"pmaddwd %%mm5, %%mm1 \n\t"
"packssdw %%mm1, %%mm0 \n\t" // V1 V0 U1 U0
"psraw $7, %%mm0 \n\t"
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
"movq 12(%0, %%"REG_b"), %%mm4 \n\t"
"movq 12(%1, %%"REG_b"), %%mm1 \n\t"
"movq 18(%0, %%"REG_b"), %%mm2 \n\t"
"movq 18(%1, %%"REG_b"), %%mm3 \n\t"
PAVGB" %%mm1, %%mm4 \n\t"
PAVGB" %%mm3, %%mm2 \n\t"
"movq %%mm4, %%mm1 \n\t"
"movq %%mm2, %%mm3 \n\t"
"psrlq $24, %%mm4 \n\t"
"psrlq $24, %%mm2 \n\t"
PAVGB" %%mm1, %%mm4 \n\t"
PAVGB" %%mm3, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm4 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
#else
"movd 12(%0, %%"REG_b"), %%mm4 \n\t"
"movd 12(%1, %%"REG_b"), %%mm1 \n\t"
"movd 15(%0, %%"REG_b"), %%mm2 \n\t"
"movd 15(%1, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm4 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"paddw %%mm1, %%mm4 \n\t"
"paddw %%mm3, %%mm2 \n\t"
"paddw %%mm2, %%mm4 \n\t"
"movd 18(%0, %%"REG_b"), %%mm5 \n\t"
"movd 18(%1, %%"REG_b"), %%mm1 \n\t"
"movd 21(%0, %%"REG_b"), %%mm2 \n\t"
"movd 21(%1, %%"REG_b"), %%mm3 \n\t"
"punpcklbw %%mm7, %%mm5 \n\t"
"punpcklbw %%mm7, %%mm1 \n\t"
"punpcklbw %%mm7, %%mm2 \n\t"
"punpcklbw %%mm7, %%mm3 \n\t"
"paddw %%mm1, %%mm5 \n\t"
"paddw %%mm3, %%mm2 \n\t"
"paddw %%mm5, %%mm2 \n\t"
"movq "MANGLE(w1111)", %%mm5 \n\t"
"psrlw $2, %%mm4 \n\t"
"psrlw $2, %%mm2 \n\t"
#endif
"movq "MANGLE(bgr2VCoeff)", %%mm1 \n\t"
"movq "MANGLE(bgr2VCoeff)", %%mm3 \n\t"
"pmaddwd %%mm4, %%mm1 \n\t"
"pmaddwd %%mm2, %%mm3 \n\t"
"pmaddwd %%mm6, %%mm4 \n\t"
"pmaddwd %%mm6, %%mm2 \n\t"
#ifndef FAST_BGR2YV12
"psrad $8, %%mm4 \n\t"
"psrad $8, %%mm1 \n\t"
"psrad $8, %%mm2 \n\t"
"psrad $8, %%mm3 \n\t"
#endif
"packssdw %%mm2, %%mm4 \n\t"
"packssdw %%mm3, %%mm1 \n\t"
"pmaddwd %%mm5, %%mm4 \n\t"
"pmaddwd %%mm5, %%mm1 \n\t"
"add $24, %%"REG_b" \n\t"
"packssdw %%mm1, %%mm4 \n\t" // V3 V2 U3 U2
"psraw $7, %%mm4 \n\t"
"movq %%mm0, %%mm1 \n\t"
"punpckldq %%mm4, %%mm0 \n\t"
"punpckhdq %%mm4, %%mm1 \n\t"
"packsswb %%mm1, %%mm0 \n\t"
"paddb "MANGLE(bgr2UVOffset)", %%mm0 \n\t"
"movd %%mm0, (%2, %%"REG_a") \n\t"
"punpckhdq %%mm0, %%mm0 \n\t"
"movd %%mm0, (%3, %%"REG_a") \n\t"
"add $4, %%"REG_a" \n\t"
" js 1b \n\t"
: : "r" (src+chromWidth*6), "r" (src+srcStride+chromWidth*6), "r" (udst+chromWidth), "r" (vdst+chromWidth), "g" (-chromWidth)
: "%"REG_a, "%"REG_b
);
udst += chromStride;
vdst += chromStride;
src += srcStride*2;
}
asm volatile( EMMS" \n\t"
SFENCE" \n\t"
:::"memory");
#else
y=0;
#endif
for(; y<height; y+=2)
{
long i;
for(i=0; i<chromWidth; i++)
{
unsigned int b= src[6*i+0];
unsigned int g= src[6*i+1];
unsigned int r= src[6*i+2];
unsigned int Y = ((RY*r + GY*g + BY*b)>>RGB2YUV_SHIFT) + 16;
unsigned int V = ((RV*r + GV*g + BV*b)>>RGB2YUV_SHIFT) + 128;
unsigned int U = ((RU*r + GU*g + BU*b)>>RGB2YUV_SHIFT) + 128;
udst[i] = U;
vdst[i] = V;
ydst[2*i] = Y;
b= src[6*i+3];
g= src[6*i+4];
r= src[6*i+5];
Y = ((RY*r + GY*g + BY*b)>>RGB2YUV_SHIFT) + 16;
ydst[2*i+1] = Y;
}
ydst += lumStride;
src += srcStride;
for(i=0; i<chromWidth; i++)
{
unsigned in⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -