summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--libswscale/colorspace-test.c22
-rw-r--r--libswscale/ppc/yuv2rgb_altivec.c36
-rw-r--r--libswscale/rgb2rgb.c16
-rw-r--r--libswscale/rgb2rgb.h16
-rw-r--r--libswscale/rgb2rgb_template.c2736
-rw-r--r--libswscale/sparc/yuv2rgb_vis.c188
-rw-r--r--libswscale/swscale.c270
-rw-r--r--libswscale/swscale_template.c228
8 files changed, 1756 insertions, 1756 deletions
diff --git a/libswscale/colorspace-test.c b/libswscale/colorspace-test.c
index 2223ee3a31..a1ec1be1b4 100644
--- a/libswscale/colorspace-test.c
+++ b/libswscale/colorspace-test.c
@@ -41,17 +41,17 @@ static char *args_parse(int argc, char *argv[])
while ((o = getopt(argc, argv, "m23")) != -1) {
switch (o) {
- case 'm':
- cpu_caps |= SWS_CPU_CAPS_MMX;
- break;
- case '2':
- cpu_caps |= SWS_CPU_CAPS_MMX2;
- break;
- case '3':
- cpu_caps |= SWS_CPU_CAPS_3DNOW;
- break;
- default:
- av_log(NULL, AV_LOG_ERROR, "Unknown option %c\n", o);
+ case 'm':
+ cpu_caps |= SWS_CPU_CAPS_MMX;
+ break;
+ case '2':
+ cpu_caps |= SWS_CPU_CAPS_MMX2;
+ break;
+ case '3':
+ cpu_caps |= SWS_CPU_CAPS_3DNOW;
+ break;
+ default:
+ av_log(NULL, AV_LOG_ERROR, "Unknown option %c\n", o);
}
}
diff --git a/libswscale/ppc/yuv2rgb_altivec.c b/libswscale/ppc/yuv2rgb_altivec.c
index 37fbcb5dd5..57acc6bc3e 100644
--- a/libswscale/ppc/yuv2rgb_altivec.c
+++ b/libswscale/ppc/yuv2rgb_altivec.c
@@ -855,13 +855,13 @@ ff_yuv2packedX_altivec(SwsContext *c,
B = vec_packclp (B0,B1);
switch(c->dstFormat) {
- case PIX_FMT_ABGR: out_abgr (R,G,B,out); break;
- case PIX_FMT_BGRA: out_bgra (R,G,B,out); break;
- case PIX_FMT_RGBA: out_rgba (R,G,B,out); break;
- case PIX_FMT_ARGB: out_argb (R,G,B,out); break;
- case PIX_FMT_RGB24: out_rgb24 (R,G,B,out); break;
- case PIX_FMT_BGR24: out_bgr24 (R,G,B,out); break;
- default:
+ case PIX_FMT_ABGR: out_abgr (R,G,B,out); break;
+ case PIX_FMT_BGRA: out_bgra (R,G,B,out); break;
+ case PIX_FMT_RGBA: out_rgba (R,G,B,out); break;
+ case PIX_FMT_ARGB: out_argb (R,G,B,out); break;
+ case PIX_FMT_RGB24: out_rgb24 (R,G,B,out); break;
+ case PIX_FMT_BGR24: out_bgr24 (R,G,B,out); break;
+ default:
{
/* If this is reached, the caller should have called yuv2packedXinC
instead. */
@@ -934,17 +934,17 @@ ff_yuv2packedX_altivec(SwsContext *c,
nout = (vector unsigned char *)scratch;
switch(c->dstFormat) {
- case PIX_FMT_ABGR: out_abgr (R,G,B,nout); break;
- case PIX_FMT_BGRA: out_bgra (R,G,B,nout); break;
- case PIX_FMT_RGBA: out_rgba (R,G,B,nout); break;
- case PIX_FMT_ARGB: out_argb (R,G,B,nout); break;
- case PIX_FMT_RGB24: out_rgb24 (R,G,B,nout); break;
- case PIX_FMT_BGR24: out_bgr24 (R,G,B,nout); break;
- default:
- /* Unreachable, I think. */
- av_log(c, AV_LOG_ERROR, "altivec_yuv2packedX doesn't support %s output\n",
- sws_format_name(c->dstFormat));
- return;
+ case PIX_FMT_ABGR: out_abgr (R,G,B,nout); break;
+ case PIX_FMT_BGRA: out_bgra (R,G,B,nout); break;
+ case PIX_FMT_RGBA: out_rgba (R,G,B,nout); break;
+ case PIX_FMT_ARGB: out_argb (R,G,B,nout); break;
+ case PIX_FMT_RGB24: out_rgb24 (R,G,B,nout); break;
+ case PIX_FMT_BGR24: out_bgr24 (R,G,B,nout); break;
+ default:
+ /* Unreachable, I think. */
+ av_log(c, AV_LOG_ERROR, "altivec_yuv2packedX doesn't support %s output\n",
+ sws_format_name(c->dstFormat));
+ return;
}
memcpy (&((uint32_t*)dest)[i], scratch, (dstW-i)/4);
diff --git a/libswscale/rgb2rgb.c b/libswscale/rgb2rgb.c
index 7aeb2a3ac2..1b1d4db77c 100644
--- a/libswscale/rgb2rgb.c
+++ b/libswscale/rgb2rgb.c
@@ -88,17 +88,17 @@ void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *sr
long srcStride1, long srcStride2,
long srcStride3, long dstStride);
void (*uyvytoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void (*uyvytoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void (*yuyvtoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
#if ARCH_X86 && CONFIG_GPL
diff --git a/libswscale/rgb2rgb.h b/libswscale/rgb2rgb.h
index 3850ef291d..9d6e6df415 100644
--- a/libswscale/rgb2rgb.h
+++ b/libswscale/rgb2rgb.h
@@ -144,17 +144,17 @@ extern void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint
extern void (*uyvytoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
extern void (*uyvytoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
extern void (*yuyvtoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
extern void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void sws_rgb2rgb_init(int flags);
diff --git a/libswscale/rgb2rgb_template.c b/libswscale/rgb2rgb_template.c
index 3d1fb8aea6..475a810f0b 100644
--- a/libswscale/rgb2rgb_template.c
+++ b/libswscale/rgb2rgb_template.c
@@ -77,16 +77,16 @@ static inline void RENAME(rgb24tobgr32)(const uint8_t *src, uint8_t *dst, long s
const uint8_t *s = src;
const uint8_t *end;
#if HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
end = s + src_size;
#if HAVE_MMX
- __asm__ volatile(PREFETCH" %0"::"m"(*s):"memory");
- mm_end = end - 23;
- __asm__ volatile("movq %0, %%mm7"::"m"(mask32a):"memory");
- while (s < mm_end)
- {
- __asm__ volatile(
+ __asm__ volatile(PREFETCH" %0"::"m"(*s):"memory");
+ mm_end = end - 23;
+ __asm__ volatile("movq %0, %%mm7"::"m"(mask32a):"memory");
+ while (s < mm_end)
+ {
+ __asm__ volatile(
PREFETCH" 32%1 \n\t"
"movd %1, %%mm0 \n\t"
"punpckldq 3%1, %%mm0 \n\t"
@@ -107,11 +107,11 @@ static inline void RENAME(rgb24tobgr32)(const uint8_t *src, uint8_t *dst, long s
:"=m"(*dest)
:"m"(*s)
:"memory");
- dest += 32;
- s += 24;
- }
- __asm__ volatile(SFENCE:::"memory");
- __asm__ volatile(EMMS:::"memory");
+ dest += 32;
+ s += 24;
+ }
+ __asm__ volatile(SFENCE:::"memory");
+ __asm__ volatile(EMMS:::"memory");
#endif
while (s < end)
{
@@ -146,53 +146,53 @@ static inline void RENAME(rgb32tobgr24)(const uint8_t *src, uint8_t *dst, long s
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movq %1, %%mm0 \n\t"
- "movq 8%1, %%mm1 \n\t"
- "movq 16%1, %%mm4 \n\t"
- "movq 24%1, %%mm5 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm1, %%mm3 \n\t"
- "movq %%mm4, %%mm6 \n\t"
- "movq %%mm5, %%mm7 \n\t"
- "psrlq $8, %%mm2 \n\t"
- "psrlq $8, %%mm3 \n\t"
- "psrlq $8, %%mm6 \n\t"
- "psrlq $8, %%mm7 \n\t"
- "pand %2, %%mm0 \n\t"
- "pand %2, %%mm1 \n\t"
- "pand %2, %%mm4 \n\t"
- "pand %2, %%mm5 \n\t"
- "pand %3, %%mm2 \n\t"
- "pand %3, %%mm3 \n\t"
- "pand %3, %%mm6 \n\t"
- "pand %3, %%mm7 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm3, %%mm1 \n\t"
- "por %%mm6, %%mm4 \n\t"
- "por %%mm7, %%mm5 \n\t"
-
- "movq %%mm1, %%mm2 \n\t"
- "movq %%mm4, %%mm3 \n\t"
- "psllq $48, %%mm2 \n\t"
- "psllq $32, %%mm3 \n\t"
- "pand %4, %%mm2 \n\t"
- "pand %5, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "psrlq $16, %%mm1 \n\t"
- "psrlq $32, %%mm4 \n\t"
- "psllq $16, %%mm5 \n\t"
- "por %%mm3, %%mm1 \n\t"
- "pand %6, %%mm5 \n\t"
- "por %%mm5, %%mm4 \n\t"
-
- MOVNTQ" %%mm0, %0 \n\t"
- MOVNTQ" %%mm1, 8%0 \n\t"
- MOVNTQ" %%mm4, 16%0"
- :"=m"(*dest)
- :"m"(*s),"m"(mask24l),
- "m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
- :"memory");
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq 8%1, %%mm1 \n\t"
+ "movq 16%1, %%mm4 \n\t"
+ "movq 24%1, %%mm5 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm1, %%mm3 \n\t"
+ "movq %%mm4, %%mm6 \n\t"
+ "movq %%mm5, %%mm7 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "psrlq $8, %%mm3 \n\t"
+ "psrlq $8, %%mm6 \n\t"
+ "psrlq $8, %%mm7 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %2, %%mm1 \n\t"
+ "pand %2, %%mm4 \n\t"
+ "pand %2, %%mm5 \n\t"
+ "pand %3, %%mm2 \n\t"
+ "pand %3, %%mm3 \n\t"
+ "pand %3, %%mm6 \n\t"
+ "pand %3, %%mm7 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm3, %%mm1 \n\t"
+ "por %%mm6, %%mm4 \n\t"
+ "por %%mm7, %%mm5 \n\t"
+
+ "movq %%mm1, %%mm2 \n\t"
+ "movq %%mm4, %%mm3 \n\t"
+ "psllq $48, %%mm2 \n\t"
+ "psllq $32, %%mm3 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "pand %5, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psrlq $16, %%mm1 \n\t"
+ "psrlq $32, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm3, %%mm1 \n\t"
+ "pand %6, %%mm5 \n\t"
+ "por %%mm5, %%mm4 \n\t"
+
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm1, 8%0 \n\t"
+ MOVNTQ" %%mm4, 16%0"
+ :"=m"(*dest)
+ :"m"(*s),"m"(mask24l),
+ "m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
+ :"memory");
dest += 24;
s += 32;
}
@@ -237,19 +237,19 @@ static inline void RENAME(rgb15to16)(const uint8_t *src, uint8_t *dst, long src_
while (s<mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movq %1, %%mm0 \n\t"
- "movq 8%1, %%mm2 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm2, %%mm3 \n\t"
- "pand %%mm4, %%mm0 \n\t"
- "pand %%mm4, %%mm2 \n\t"
- "paddw %%mm1, %%mm0 \n\t"
- "paddw %%mm3, %%mm2 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- MOVNTQ" %%mm2, 8%0"
- :"=m"(*d)
- :"m"(*s)
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "pand %%mm4, %%mm0 \n\t"
+ "pand %%mm4, %%mm2 \n\t"
+ "paddw %%mm1, %%mm0 \n\t"
+ "paddw %%mm3, %%mm2 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm2, 8%0"
+ :"=m"(*d)
+ :"m"(*s)
);
d+=16;
s+=16;
@@ -287,23 +287,23 @@ static inline void RENAME(rgb16to15)(const uint8_t *src, uint8_t *dst, long src_
while (s<mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movq %1, %%mm0 \n\t"
- "movq 8%1, %%mm2 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm2, %%mm3 \n\t"
- "psrlq $1, %%mm0 \n\t"
- "psrlq $1, %%mm2 \n\t"
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm7, %%mm2 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm3 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm3, %%mm2 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- MOVNTQ" %%mm2, 8%0"
- :"=m"(*d)
- :"m"(*s)
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "psrlq $1, %%mm0 \n\t"
+ "psrlq $1, %%mm2 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm3 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm3, %%mm2 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm2, 8%0"
+ :"=m"(*d)
+ :"m"(*s)
);
d+=16;
s+=16;
@@ -339,38 +339,38 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, long src_
mm_end = end - 15;
#if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
__asm__ volatile(
- "movq %3, %%mm5 \n\t"
- "movq %4, %%mm6 \n\t"
- "movq %5, %%mm7 \n\t"
- "jmp 2f \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1) \n\t"
- "movd (%1), %%mm0 \n\t"
- "movd 4(%1), %%mm3 \n\t"
- "punpckldq 8(%1), %%mm0 \n\t"
- "punpckldq 12(%1), %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "pand %%mm6, %%mm0 \n\t"
- "pand %%mm6, %%mm3 \n\t"
- "pmaddwd %%mm7, %%mm0 \n\t"
- "pmaddwd %%mm7, %%mm3 \n\t"
- "pand %%mm5, %%mm1 \n\t"
- "pand %%mm5, %%mm4 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "psrld $5, %%mm0 \n\t"
- "pslld $11, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, (%0) \n\t"
- "add $16, %1 \n\t"
- "add $8, %0 \n\t"
- "2: \n\t"
- "cmp %2, %1 \n\t"
- " jb 1b \n\t"
- : "+r" (d), "+r"(s)
- : "r" (mm_end), "m" (mask3216g), "m" (mask3216br), "m" (mul3216)
+ "movq %3, %%mm5 \n\t"
+ "movq %4, %%mm6 \n\t"
+ "movq %5, %%mm7 \n\t"
+ "jmp 2f \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1) \n\t"
+ "movd (%1), %%mm0 \n\t"
+ "movd 4(%1), %%mm3 \n\t"
+ "punpckldq 8(%1), %%mm0 \n\t"
+ "punpckldq 12(%1), %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "pand %%mm6, %%mm0 \n\t"
+ "pand %%mm6, %%mm3 \n\t"
+ "pmaddwd %%mm7, %%mm0 \n\t"
+ "pmaddwd %%mm7, %%mm3 \n\t"
+ "pand %%mm5, %%mm1 \n\t"
+ "pand %%mm5, %%mm4 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "psrld $5, %%mm0 \n\t"
+ "pslld $11, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, (%0) \n\t"
+ "add $16, %1 \n\t"
+ "add $8, %0 \n\t"
+ "2: \n\t"
+ "cmp %2, %1 \n\t"
+ " jb 1b \n\t"
+ : "+r" (d), "+r"(s)
+ : "r" (mm_end), "m" (mask3216g), "m" (mask3216br), "m" (mul3216)
);
#else
__asm__ volatile(PREFETCH" %0"::"m"(*src):"memory");
@@ -381,35 +381,35 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, long src_
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movd %1, %%mm0 \n\t"
- "movd 4%1, %%mm3 \n\t"
- "punpckldq 8%1, %%mm0 \n\t"
- "punpckldq 12%1, %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "movq %%mm3, %%mm5 \n\t"
- "psrlq $3, %%mm0 \n\t"
- "psrlq $3, %%mm3 \n\t"
- "pand %2, %%mm0 \n\t"
- "pand %2, %%mm3 \n\t"
- "psrlq $5, %%mm1 \n\t"
- "psrlq $5, %%mm4 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "psrlq $8, %%mm2 \n\t"
- "psrlq $8, %%mm5 \n\t"
- "pand %%mm7, %%mm2 \n\t"
- "pand %%mm7, %%mm5 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm5, %%mm3 \n\t"
- "psllq $16, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "movd 4%1, %%mm3 \n\t"
+ "punpckldq 8%1, %%mm0 \n\t"
+ "punpckldq 12%1, %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "psrlq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm3 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %2, %%mm3 \n\t"
+ "psrlq $5, %%mm1 \n\t"
+ "psrlq $5, %%mm4 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "psrlq $8, %%mm5 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "pand %%mm7, %%mm5 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ "psllq $16, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
d += 4;
s += 16;
}
@@ -443,35 +443,35 @@ static inline void RENAME(rgb32tobgr16)(const uint8_t *src, uint8_t *dst, long s
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movd %1, %%mm0 \n\t"
- "movd 4%1, %%mm3 \n\t"
- "punpckldq 8%1, %%mm0 \n\t"
- "punpckldq 12%1, %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "movq %%mm3, %%mm5 \n\t"
- "psllq $8, %%mm0 \n\t"
- "psllq $8, %%mm3 \n\t"
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm7, %%mm3 \n\t"
- "psrlq $5, %%mm1 \n\t"
- "psrlq $5, %%mm4 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "psrlq $19, %%mm2 \n\t"
- "psrlq $19, %%mm5 \n\t"
- "pand %2, %%mm2 \n\t"
- "pand %2, %%mm5 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm5, %%mm3 \n\t"
- "psllq $16, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "movd 4%1, %%mm3 \n\t"
+ "punpckldq 8%1, %%mm0 \n\t"
+ "punpckldq 12%1, %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "psllq $8, %%mm0 \n\t"
+ "psllq $8, %%mm3 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm7, %%mm3 \n\t"
+ "psrlq $5, %%mm1 \n\t"
+ "psrlq $5, %%mm4 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "psrlq $19, %%mm2 \n\t"
+ "psrlq $19, %%mm5 \n\t"
+ "pand %2, %%mm2 \n\t"
+ "pand %2, %%mm5 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ "psllq $16, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
d += 4;
s += 16;
}
@@ -498,38 +498,38 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, long src_
mm_end = end - 15;
#if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
__asm__ volatile(
- "movq %3, %%mm5 \n\t"
- "movq %4, %%mm6 \n\t"
- "movq %5, %%mm7 \n\t"
- "jmp 2f \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1) \n\t"
- "movd (%1), %%mm0 \n\t"
- "movd 4(%1), %%mm3 \n\t"
- "punpckldq 8(%1), %%mm0 \n\t"
- "punpckldq 12(%1), %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "pand %%mm6, %%mm0 \n\t"
- "pand %%mm6, %%mm3 \n\t"
- "pmaddwd %%mm7, %%mm0 \n\t"
- "pmaddwd %%mm7, %%mm3 \n\t"
- "pand %%mm5, %%mm1 \n\t"
- "pand %%mm5, %%mm4 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "psrld $6, %%mm0 \n\t"
- "pslld $10, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, (%0) \n\t"
- "add $16, %1 \n\t"
- "add $8, %0 \n\t"
- "2: \n\t"
- "cmp %2, %1 \n\t"
- " jb 1b \n\t"
- : "+r" (d), "+r"(s)
- : "r" (mm_end), "m" (mask3215g), "m" (mask3216br), "m" (mul3215)
+ "movq %3, %%mm5 \n\t"
+ "movq %4, %%mm6 \n\t"
+ "movq %5, %%mm7 \n\t"
+ "jmp 2f \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1) \n\t"
+ "movd (%1), %%mm0 \n\t"
+ "movd 4(%1), %%mm3 \n\t"
+ "punpckldq 8(%1), %%mm0 \n\t"
+ "punpckldq 12(%1), %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "pand %%mm6, %%mm0 \n\t"
+ "pand %%mm6, %%mm3 \n\t"
+ "pmaddwd %%mm7, %%mm0 \n\t"
+ "pmaddwd %%mm7, %%mm3 \n\t"
+ "pand %%mm5, %%mm1 \n\t"
+ "pand %%mm5, %%mm4 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "psrld $6, %%mm0 \n\t"
+ "pslld $10, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, (%0) \n\t"
+ "add $16, %1 \n\t"
+ "add $8, %0 \n\t"
+ "2: \n\t"
+ "cmp %2, %1 \n\t"
+ " jb 1b \n\t"
+ : "+r" (d), "+r"(s)
+ : "r" (mm_end), "m" (mask3215g), "m" (mask3216br), "m" (mul3215)
);
#else
__asm__ volatile(PREFETCH" %0"::"m"(*src):"memory");
@@ -540,35 +540,35 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, long src_
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movd %1, %%mm0 \n\t"
- "movd 4%1, %%mm3 \n\t"
- "punpckldq 8%1, %%mm0 \n\t"
- "punpckldq 12%1, %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "movq %%mm3, %%mm5 \n\t"
- "psrlq $3, %%mm0 \n\t"
- "psrlq $3, %%mm3 \n\t"
- "pand %2, %%mm0 \n\t"
- "pand %2, %%mm3 \n\t"
- "psrlq $6, %%mm1 \n\t"
- "psrlq $6, %%mm4 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "psrlq $9, %%mm2 \n\t"
- "psrlq $9, %%mm5 \n\t"
- "pand %%mm7, %%mm2 \n\t"
- "pand %%mm7, %%mm5 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm5, %%mm3 \n\t"
- "psllq $16, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "movd 4%1, %%mm3 \n\t"
+ "punpckldq 8%1, %%mm0 \n\t"
+ "punpckldq 12%1, %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "psrlq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm3 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %2, %%mm3 \n\t"
+ "psrlq $6, %%mm1 \n\t"
+ "psrlq $6, %%mm4 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "psrlq $9, %%mm2 \n\t"
+ "psrlq $9, %%mm5 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "pand %%mm7, %%mm5 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ "psllq $16, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
d += 4;
s += 16;
}
@@ -602,35 +602,35 @@ static inline void RENAME(rgb32tobgr15)(const uint8_t *src, uint8_t *dst, long s
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movd %1, %%mm0 \n\t"
- "movd 4%1, %%mm3 \n\t"
- "punpckldq 8%1, %%mm0 \n\t"
- "punpckldq 12%1, %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "movq %%mm3, %%mm5 \n\t"
- "psllq $7, %%mm0 \n\t"
- "psllq $7, %%mm3 \n\t"
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm7, %%mm3 \n\t"
- "psrlq $6, %%mm1 \n\t"
- "psrlq $6, %%mm4 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "psrlq $19, %%mm2 \n\t"
- "psrlq $19, %%mm5 \n\t"
- "pand %2, %%mm2 \n\t"
- "pand %2, %%mm5 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm5, %%mm3 \n\t"
- "psllq $16, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "movd 4%1, %%mm3 \n\t"
+ "punpckldq 8%1, %%mm0 \n\t"
+ "punpckldq 12%1, %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "psllq $7, %%mm0 \n\t"
+ "psllq $7, %%mm3 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm7, %%mm3 \n\t"
+ "psrlq $6, %%mm1 \n\t"
+ "psrlq $6, %%mm4 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "psrlq $19, %%mm2 \n\t"
+ "psrlq $19, %%mm5 \n\t"
+ "pand %2, %%mm2 \n\t"
+ "pand %2, %%mm5 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ "psllq $16, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
d += 4;
s += 16;
}
@@ -663,35 +663,35 @@ static inline void RENAME(rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long s
while (s < mm_end)
{
__asm__ volatile(
- PREFETCH" 32%1 \n\t"
- "movd %1, %%mm0 \n\t"
- "movd 3%1, %%mm3 \n\t"
- "punpckldq 6%1, %%mm0 \n\t"
- "punpckldq 9%1, %%mm3 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm3, %%mm4 \n\t"
- "movq %%mm3, %%mm5 \n\t"
- "psrlq $3, %%mm0 \n\t"
- "psrlq $3, %%mm3 \n\t"
- "pand %2, %%mm0 \n\t"
- "pand %2, %%mm3 \n\t"
- "psrlq $5, %%mm1 \n\t"
- "psrlq $5, %%mm4 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "psrlq $8, %%mm2 \n\t"
- "psrlq $8, %%mm5 \n\t"
- "pand %%mm7, %%mm2 \n\t"
- "pand %%mm7, %%mm5 \n\t"
- "por %%mm1, %%mm0 \n\t"
- "por %%mm4, %%mm3 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm5, %%mm3 \n\t"
- "psllq $16, %%mm3 \n\t"
- "por %%mm3, %%mm0 \n\t"
- MOVNTQ" %%mm0, %0 \n\t"
- :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "movd 3%1, %%mm3 \n\t"
+ "punpckldq 6%1, %%mm0 \n\t"
+ "punpckldq 9%1, %%mm3 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm3, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "psrlq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm3 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %2, %%mm3 \n\t"
+ "psrlq $5, %%mm1 \n\t"
+ "psrlq $5, %%mm4 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "psrlq $8, %%mm5 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "pand %%mm7, %%mm5 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ "psllq $16, %%mm3 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
d += 4;
s += 12;
}
@@ -726,35 +726,35 @@ st