aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorDiego Biurrun <diego@biurrun.de>2007-04-28 11:44:49 +0000
committerDiego Biurrun <diego@biurrun.de>2007-04-28 11:44:49 +0000
commit6e42e6c4b410dbef8b593c2d796a5dad95f89ee4 (patch)
treea85ea55f0fcc2a85fb2001b954df839cd5b715d1
parent1da87823db5dbeed72a50fd0194f9397052a5130 (diff)
downloadffmpeg-6e42e6c4b410dbef8b593c2d796a5dad95f89ee4.tar.gz
cosmetics attack, part I: Remove all tabs and prettyprint/reindent the code.
Originally committed as revision 23158 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale
Diffstat
-rw-r--r--libswscale/cs_test.c6
-rw-r--r--libswscale/rgb2rgb.c609
-rw-r--r--libswscale/rgb2rgb.h69
-rw-r--r--libswscale/rgb2rgb_template.c4619
-rw-r--r--libswscale/swscale_altivec_template.c852
5 files changed, 3074 insertions, 3081 deletions
diff --git a/libswscale/cs_test.c b/libswscale/cs_test.c
index 552fafaf6a..876f270411 100644
--- a/libswscale/cs_test.c
+++ b/libswscale/cs_test.c
@@ -142,7 +142,7 @@ int main(int argc, char **argv)
for(i=0; i<SIZE; i++){
if(srcBuffer[i]!=srcByte){
av_log(NULL, AV_LOG_INFO, "src damaged at %d w:%d src:%d dst:%d %s\n",
- i, width, srcOffset, dstOffset, name);
+ i, width, srcOffset, dstOffset, name);
failed=1;
break;
}
@@ -150,7 +150,7 @@ int main(int argc, char **argv)
for(i=0; i<dstOffset; i++){
if(dstBuffer[i]!=dstByte){
av_log(NULL, AV_LOG_INFO, "dst damaged at %d w:%d src:%d dst:%d %s\n",
- i, width, srcOffset, dstOffset, name);
+ i, width, srcOffset, dstOffset, name);
failed=1;
break;
}
@@ -158,7 +158,7 @@ int main(int argc, char **argv)
for(i=dstOffset + width*dstBpp; i<SIZE; i++){
if(dstBuffer[i]!=dstByte){
av_log(NULL, AV_LOG_INFO, "dst damaged at %d w:%d src:%d dst:%d %s\n",
- i, width, srcOffset, dstOffset, name);
+ i, width, srcOffset, dstOffset, name);
failed=1;
break;
}
diff --git a/libswscale/rgb2rgb.c b/libswscale/rgb2rgb.c
index 9eadbe095c..d363a2857b 100644
--- a/libswscale/rgb2rgb.c
+++ b/libswscale/rgb2rgb.c
@@ -1,5 +1,4 @@
/*
- *
* rgb2rgb.c, Software RGB to RGB convertor
* pluralize by Software PAL8 to RGB convertor
* Software YUV to YUV convertor
@@ -61,83 +60,83 @@ void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
void (*planar2x)(const uint8_t *src, uint8_t *dst, long width, long height,
- long srcStride, long dstStride);
+ long srcStride, long dstStride);
void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
- long width, long height, long src1Stride,
- long src2Stride, long dstStride);
+ long width, long height, long src1Stride,
+ long src2Stride, long dstStride);
void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
- uint8_t *dst1, uint8_t *dst2,
- long width, long height,
- long srcStride1, long srcStride2,
- long dstStride1, long dstStride2);
+ uint8_t *dst1, uint8_t *dst2,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long dstStride1, long dstStride2);
void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
- uint8_t *dst,
- long width, long height,
- long srcStride1, long srcStride2,
- long srcStride3, long dstStride);
+ uint8_t *dst,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long srcStride3, long dstStride);
#if defined(ARCH_X86) && defined(CONFIG_GPL)
-static const uint64_t mmx_null __attribute__((aligned(8))) = 0x0000000000000000ULL;
-static const uint64_t mmx_one __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
-static const uint64_t mask32b attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
-static const uint64_t mask32g attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
-static const uint64_t mask32r attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
-static const uint64_t mask32 __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
-static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
-static const uint64_t mask3216g __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
-static const uint64_t mask3215g __attribute__((aligned(8)))=0x0000F8000000F800ULL;
-static const uint64_t mul3216 __attribute__((aligned(8))) = 0x2000000420000004ULL;
-static const uint64_t mul3215 __attribute__((aligned(8))) = 0x2000000820000008ULL;
-static const uint64_t mask24b attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
-static const uint64_t mask24g attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
-static const uint64_t mask24r attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
-static const uint64_t mask24l __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
-static const uint64_t mask24h __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
-static const uint64_t mask24hh __attribute__((aligned(8))) = 0xffff000000000000ULL;
-static const uint64_t mask24hhh __attribute__((aligned(8))) = 0xffffffff00000000ULL;
-static const uint64_t mask24hhhh __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
-static const uint64_t mask15b __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
-static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
-static const uint64_t mask15s __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
-static const uint64_t mask15g __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
-static const uint64_t mask15r __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
+static const uint64_t mmx_null __attribute__((aligned(8))) = 0x0000000000000000ULL;
+static const uint64_t mmx_one __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
+static const uint64_t mask32b attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
+static const uint64_t mask32g attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
+static const uint64_t mask32r attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
+static const uint64_t mask32 __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
+static const uint64_t mask3216br __attribute__((aligned(8))) = 0x00F800F800F800F8ULL;
+static const uint64_t mask3216g __attribute__((aligned(8))) = 0x0000FC000000FC00ULL;
+static const uint64_t mask3215g __attribute__((aligned(8))) = 0x0000F8000000F800ULL;
+static const uint64_t mul3216 __attribute__((aligned(8))) = 0x2000000420000004ULL;
+static const uint64_t mul3215 __attribute__((aligned(8))) = 0x2000000820000008ULL;
+static const uint64_t mask24b attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
+static const uint64_t mask24g attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
+static const uint64_t mask24r attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
+static const uint64_t mask24l __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
+static const uint64_t mask24h __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
+static const uint64_t mask24hh __attribute__((aligned(8))) = 0xffff000000000000ULL;
+static const uint64_t mask24hhh __attribute__((aligned(8))) = 0xffffffff00000000ULL;
+static const uint64_t mask24hhhh __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
+static const uint64_t mask15b __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
+static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
+static const uint64_t mask15s __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
+static const uint64_t mask15g __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
+static const uint64_t mask15r __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
-static const uint64_t mask16g __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
-static const uint64_t mask16r __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
-static const uint64_t red_16mask __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
-static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
-static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
-static const uint64_t red_15mask __attribute__((aligned(8))) = 0x00007c0000007c00ULL;
-static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000003e0ULL;
-static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
+static const uint64_t mask16g __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
+static const uint64_t mask16r __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
+static const uint64_t red_16mask __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
+static const uint64_t green_16mask __attribute__((aligned(8))) = 0x000007e0000007e0ULL;
+static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
+static const uint64_t red_15mask __attribute__((aligned(8))) = 0x00007c0000007c00ULL;
+static const uint64_t green_15mask __attribute__((aligned(8))) = 0x000003e0000003e0ULL;
+static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
#ifdef FAST_BGR2YV12
-static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
-static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
-static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
+static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
+static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
+static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
#else
-static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
-static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
-static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
+static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
+static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
+static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
#endif
-static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
-static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL;
-static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;
+static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
+static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8))) = 0x8080808080808080ULL;
+static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;
#if 0
static volatile uint64_t __attribute__((aligned(8))) b5Dither;
@@ -146,12 +145,12 @@ static volatile uint64_t __attribute__((aligned(8))) g6Dither;
static volatile uint64_t __attribute__((aligned(8))) r5Dither;
static uint64_t __attribute__((aligned(8))) dither4[2]={
- 0x0103010301030103LL,
- 0x0200020002000200LL,};
+ 0x0103010301030103LL,
+ 0x0200020002000200LL,};
static uint64_t __attribute__((aligned(8))) dither8[2]={
- 0x0602060206020602LL,
- 0x0004000400040004LL,};
+ 0x0602060206020602LL,
+ 0x0004000400040004LL,};
#endif
#endif /* defined(ARCH_X86) */
@@ -215,15 +214,15 @@ static uint64_t __attribute__((aligned(8))) dither8[2]={
void sws_rgb2rgb_init(int flags){
#if (defined(HAVE_MMX2) || defined(HAVE_3DNOW) || defined(HAVE_MMX)) && defined(CONFIG_GPL)
- if(flags & SWS_CPU_CAPS_MMX2)
- rgb2rgb_init_MMX2();
- else if(flags & SWS_CPU_CAPS_3DNOW)
- rgb2rgb_init_3DNOW();
- else if(flags & SWS_CPU_CAPS_MMX)
- rgb2rgb_init_MMX();
- else
+ if (flags & SWS_CPU_CAPS_MMX2)
+ rgb2rgb_init_MMX2();
+ else if (flags & SWS_CPU_CAPS_3DNOW)
+ rgb2rgb_init_3DNOW();
+ else if (flags & SWS_CPU_CAPS_MMX)
+ rgb2rgb_init_MMX();
+ else
#endif /* defined(HAVE_MMX2) || defined(HAVE_3DNOW) || defined(HAVE_MMX) */
- rgb2rgb_init_C();
+ rgb2rgb_init_C();
}
/**
@@ -231,49 +230,49 @@ void sws_rgb2rgb_init(int flags){
*/
void palette8torgb32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
+ long i;
/*
- for(i=0; i<num_pixels; i++)
- ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
+ for (i=0; i<num_pixels; i++)
+ ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
*/
- for(i=0; i<num_pixels; i++)
- {
- #ifdef WORDS_BIGENDIAN
- dst[3]= palette[ src[i]*4+2 ];
- dst[2]= palette[ src[i]*4+1 ];
- dst[1]= palette[ src[i]*4+0 ];
- #else
- //FIXME slow?
- dst[0]= palette[ src[i]*4+2 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+0 ];
- //dst[3]= 0; /* do we need this cleansing? */
- #endif
- dst+= 4;
- }
+ for (i=0; i<num_pixels; i++)
+ {
+ #ifdef WORDS_BIGENDIAN
+ dst[3]= palette[ src[i]*4+2 ];
+ dst[2]= palette[ src[i]*4+1 ];
+ dst[1]= palette[ src[i]*4+0 ];
+ #else
+ //FIXME slow?
+ dst[0]= palette[ src[i]*4+2 ];
+ dst[1]= palette[ src[i]*4+1 ];
+ dst[2]= palette[ src[i]*4+0 ];
+ //dst[3]= 0; /* do we need this cleansing? */
+ #endif
+ dst+= 4;
+ }
}
void palette8tobgr32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
- for(i=0; i<num_pixels; i++)
- {
- #ifdef WORDS_BIGENDIAN
- dst[3]= palette[ src[i]*4+0 ];
- dst[2]= palette[ src[i]*4+1 ];
- dst[1]= palette[ src[i]*4+2 ];
- #else
- //FIXME slow?
- dst[0]= palette[ src[i]*4+0 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+2 ];
- //dst[3]= 0; /* do we need this cleansing? */
- #endif
-
- dst+= 4;
- }
+ long i;
+ for (i=0; i<num_pixels; i++)
+ {
+ #ifdef WORDS_BIGENDIAN
+ dst[3]= palette[ src[i]*4+0 ];
+ dst[2]= palette[ src[i]*4+1 ];
+ dst[1]= palette[ src[i]*4+2 ];
+ #else
+ //FIXME slow?
+ dst[0]= palette[ src[i]*4+0 ];
+ dst[1]= palette[ src[i]*4+1 ];
+ dst[2]= palette[ src[i]*4+2 ];
+ //dst[3]= 0; /* do we need this cleansing? */
+ #endif
+
+ dst+= 4;
+ }
}
/**
@@ -281,38 +280,38 @@ void palette8tobgr32(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
*/
void palette8torgb24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
+ long i;
/*
- writes 1 byte o much and might cause alignment issues on some architectures?
- for(i=0; i<num_pixels; i++)
- ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
+ writes 1 byte o much and might cause alignment issues on some architectures?
+ for (i=0; i<num_pixels; i++)
+ ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
*/
- for(i=0; i<num_pixels; i++)
- {
- //FIXME slow?
- dst[0]= palette[ src[i]*4+2 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+0 ];
- dst+= 3;
- }
+ for (i=0; i<num_pixels; i++)
+ {
+ //FIXME slow?
+ dst[0]= palette[ src[i]*4+2 ];
+ dst[1]= palette[ src[i]*4+1 ];
+ dst[2]= palette[ src[i]*4+0 ];
+ dst+= 3;
+ }
}
void palette8tobgr24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
+ long i;
/*
- writes 1 byte o much and might cause alignment issues on some architectures?
- for(i=0; i<num_pixels; i++)
- ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
+ writes 1 byte o much and might cause alignment issues on some architectures?
+ for (i=0; i<num_pixels; i++)
+ ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
*/
- for(i=0; i<num_pixels; i++)
- {
- //FIXME slow?
- dst[0]= palette[ src[i]*4+0 ];
- dst[1]= palette[ src[i]*4+1 ];
- dst[2]= palette[ src[i]*4+2 ];
- dst+= 3;
- }
+ for (i=0; i<num_pixels; i++)
+ {
+ //FIXME slow?
+ dst[0]= palette[ src[i]*4+0 ];
+ dst[1]= palette[ src[i]*4+1 ];
+ dst[2]= palette[ src[i]*4+2 ];
+ dst+= 3;
+ }
}
/**
@@ -320,15 +319,15 @@ void palette8tobgr24(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
*/
void palette8torgb16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
- for(i=0; i<num_pixels; i++)
- ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
+ long i;
+ for (i=0; i<num_pixels; i++)
+ ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
}
void palette8tobgr16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
- for(i=0; i<num_pixels; i++)
- ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
+ long i;
+ for (i=0; i<num_pixels; i++)
+ ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
}
/**
@@ -336,216 +335,216 @@ void palette8tobgr16(const uint8_t *src, uint8_t *dst, long num_pixels, const ui
*/
void palette8torgb15(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
- for(i=0; i<num_pixels; i++)
- ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
+ long i;
+ for (i=0; i<num_pixels; i++)
+ ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
}
void palette8tobgr15(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
{
- long i;
- for(i=0; i<num_pixels; i++)
- ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
+ long i;
+ for (i=0; i<num_pixels; i++)
+ ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
}
void rgb32tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size >> 2;
- for(i=0; i<num_pixels; i++)
- {
- #ifdef WORDS_BIGENDIAN
- /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
- dst[3*i + 0] = src[4*i + 1];
- dst[3*i + 1] = src[4*i + 2];
- dst[3*i + 2] = src[4*i + 3];
- #else
- dst[3*i + 0] = src[4*i + 2];
- dst[3*i + 1] = src[4*i + 1];
- dst[3*i + 2] = src[4*i + 0];
- #endif
- }
+ long i;
+ long num_pixels = src_size >> 2;
+ for (i=0; i<num_pixels; i++)
+ {
+ #ifdef WORDS_BIGENDIAN
+ /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
+ dst[3*i + 0] = src[4*i + 1];
+ dst[3*i + 1] = src[4*i + 2];
+ dst[3*i + 2] = src[4*i + 3];
+ #else
+ dst[3*i + 0] = src[4*i + 2];
+ dst[3*i + 1] = src[4*i + 1];
+ dst[3*i + 2] = src[4*i + 0];
+ #endif
+ }
}
void rgb24tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- for(i=0; 3*i<src_size; i++)
- {
- #ifdef WORDS_BIGENDIAN
- /* RGB24 (= R,G,B) -> BGR32 (= A,R,G,B) */
- dst[4*i + 0] = 0;
- dst[4*i + 1] = src[3*i + 0];
- dst[4*i + 2] = src[3*i + 1];
- dst[4*i + 3] = src[3*i + 2];
- #else
- dst[4*i + 0] = src[3*i + 2];
- dst[4*i + 1] = src[3*i + 1];
- dst[4*i + 2] = src[3*i + 0];
- dst[4*i + 3] = 0;
- #endif
- }
+ long i;
+ for (i=0; 3*i<src_size; i++)
+ {
+ #ifdef WORDS_BIGENDIAN
+ /* RGB24 (= R,G,B) -> BGR32 (= A,R,G,B) */
+ dst[4*i + 0] = 0;
+ dst[4*i + 1] = src[3*i + 0];
+ dst[4*i + 2] = src[3*i + 1];
+ dst[4*i + 3] = src[3*i + 2];
+ #else
+ dst[4*i + 0] = src[3*i + 2];
+ dst[4*i + 1] = src[3*i + 1];
+ dst[4*i + 2] = src[3*i + 0];
+ dst[4*i + 3] = 0;
+ #endif
+ }
}
void rgb16tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (uint16_t *)src;
- end = s + src_size/2;
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- #ifdef WORDS_BIGENDIAN
- *d++ = 0;
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0xF800)>>8;
- #else
- *d++ = (bgr&0xF800)>>8;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0x1F)<<3;
- *d++ = 0;
- #endif
- }
+ const uint16_t *end;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (uint16_t *)src;
+ end = s + src_size/2;
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ #ifdef WORDS_BIGENDIAN
+ *d++ = 0;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0xF800)>>8;
+ #else
+ *d++ = (bgr&0xF800)>>8;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = 0;
+ #endif
+ }
}
void rgb16tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (const uint16_t *)src;
- end = s + src_size/2;
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- *d++ = (bgr&0xF800)>>8;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0x1F)<<3;
- }
+ const uint16_t *end;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (const uint16_t *)src;
+ end = s + src_size/2;
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ *d++ = (bgr&0xF800)>>8;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0x1F)<<3;
+ }
}
void rgb16tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size >> 1;
-
- for(i=0; i<num_pixels; i++)
- {
- unsigned b,g,r;
- register uint16_t rgb;
- rgb = src[2*i];
- r = rgb&0x1F;
- g = (rgb&0x7E0)>>5;
- b = (rgb&0xF800)>>11;
- dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
- }
+ long i;
+ long num_pixels = src_size >> 1;
+
+ for (i=0; i<num_pixels; i++)
+ {
+ unsigned b,g,r;
+ register uint16_t rgb;
+ rgb = src[2*i];
+ r = rgb&0x1F;
+ g = (rgb&0x7E0)>>5;
+ b = (rgb&0xF800)>>11;
+ dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
+ }
}
void rgb16tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size >> 1;
-
- for(i=0; i<num_pixels; i++)
- {
- unsigned b,g,r;
- register uint16_t rgb;
- rgb = src[2*i];
- r = rgb&0x1F;
- g = (rgb&0x7E0)>>5;
- b = (rgb&0xF800)>>11;
- dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
- }
+ long i;
+ long num_pixels = src_size >> 1;
+
+ for (i=0; i<num_pixels; i++)
+ {
+ unsigned b,g,r;
+ register uint16_t rgb;
+ rgb = src[2*i];
+ r = rgb&0x1F;
+ g = (rgb&0x7E0)>>5;
+ b = (rgb&0xF800)>>11;
+ dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
+ }
}
void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (const uint16_t *)src;
- end = s + src_size/2;
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- #ifdef WORDS_BIGENDIAN
- *d++ = 0;
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x7C00)>>7;
- #else
- *d++ = (bgr&0x7C00)>>7;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x1F)<<3;
- *d++ = 0;
- #endif
- }
+ const uint16_t *end;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (const uint16_t *)src;
+ end = s + src_size/2;
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ #ifdef WORDS_BIGENDIAN
+ *d++ = 0;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x7C00)>>7;
+ #else
+ *d++ = (bgr&0x7C00)>>7;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = 0;
+ #endif
+ }
}
void rgb15tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (uint16_t *)src;
- end = s + src_size/2;
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- *d++ = (bgr&0x7C00)>>7;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x1F)<<3;
- }
+ const uint16_t *end;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (uint16_t *)src;
+ end = s + src_size/2;
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ *d++ = (bgr&0x7C00)>>7;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x1F)<<3;
+ }
}
void rgb15tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size >> 1;
-
- for(i=0; i<num_pixels; i++)
- {
- unsigned b,g,r;
- register uint16_t rgb;
- rgb = src[2*i];
- r = rgb&0x1F;
- g = (rgb&0x3E0)>>5;
- b = (rgb&0x7C00)>>10;
- dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
- }
+ long i;
+ long num_pixels = src_size >> 1;
+
+ for (i=0; i<num_pixels; i++)
+ {
+ unsigned b,g,r;
+ register uint16_t rgb;
+ rgb = src[2*i];
+ r = rgb&0x1F;
+ g = (rgb&0x3E0)>>5;
+ b = (rgb&0x7C00)>>10;
+ dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
+ }
}
void rgb15tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size >> 1;
-
- for(i=0; i<num_pixels; i++)
- {
- unsigned b,g,r;
- register uint16_t rgb;
- rgb = src[2*i];
- r = rgb&0x1F;
- g = (rgb&0x3E0)>>5;
- b = (rgb&0x7C00)>>10;
- dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
- }
+ long i;
+ long num_pixels = src_size >> 1;
+
+ for (i=0; i<num_pixels; i++)
+ {
+ unsigned b,g,r;
+ register uint16_t rgb;
+ rgb = src[2*i];
+ r = rgb&0x1F;
+ g = (rgb&0x3E0)>>5;
+ b = (rgb&0x7C00)>>10;
+ dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
+ }
}
void rgb8tobgr8(const uint8_t *src, uint8_t *dst, long src_size)
{
- long i;
- long num_pixels = src_size;
- for(i=0; i<num_pixels; i++)
- {
- unsigned b,g,r;
- register uint8_t rgb;
- rgb = src[i];
- r = (rgb&0x07);
- g = (rgb&0x38)>>3;
- b = (rgb&0xC0)>>6;
- dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
- }
+ long i;
+ long num_pixels = src_size;
+ for (i=0; i<num_pixels; i++)
+ {
+ unsigned b,g,r;
+ register uint8_t rgb;
+ rgb = src[i];
+ r = (rgb&0x07);
+ g = (rgb&0x38)>>3;
+ b = (rgb&0xC0)>>6;
+ dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
+ }
}
diff --git a/libswscale/rgb2rgb.h b/libswscale/rgb2rgb.h
index 769811fb8f..5ad3ec71b4 100644
--- a/libswscale/rgb2rgb.h
+++ b/libswscale/rgb2rgb.h
@@ -1,5 +1,4 @@
/*
- *
* rgb2rgb.h, Software RGB to RGB convertor
* pluralize by Software PAL8 to RGB convertor
* Software YUV to YUV convertor
@@ -30,18 +29,18 @@
// Note: do not fix the dependence on stdio.h
/* A full collection of rgb to rgb(bgr) convertors */
-extern void (*rgb24to32)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb24to16)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb24to15)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb32to24)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb32to16)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb32to15)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb15to16)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb15to24)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb15to32)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb16to15)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb16to24)(const uint8_t *src,uint8_t *dst,long src_size);
-extern void (*rgb16to32)(const uint8_t *src,uint8_t *dst,long src_size);
+extern void (*rgb24to32) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb24to16) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb24to15) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb32to24) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb32to16) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb32to15) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb15to16) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb15to24) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb15to32) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb16to15) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb16to24) (const uint8_t *src, uint8_t *dst, long src_size);
+extern void (*rgb16to32) (const uint8_t *src, uint8_t *dst, long src_size);
extern void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
extern void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
extern void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
@@ -59,7 +58,7 @@ extern void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size);
extern void rgb15tobgr24(const uint8_t *src, uint8_t *dst, long src_size);
extern void rgb15tobgr16(const uint8_t *src, uint8_t *dst, long src_size);
extern void rgb15tobgr15(const uint8_t *src, uint8_t *dst, long src_size);
-extern void rgb8tobgr8(const uint8_t *src, uint8_t *dst, long src_size);
+extern void rgb8tobgr8 (const uint8_t *src, uint8_t *dst, long src_size);
extern void palette8torgb32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette);
@@ -85,16 +84,16 @@ extern void palette8tobgr15(const uint8_t *src, uint8_t *dst, long num_pixels, c
* problem for anyone then tell me, and ill fix it)
*/
extern void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
/**
*
* width should be a multiple of 16
*/
extern void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
/**
*
@@ -102,8 +101,8 @@ extern void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uin
* problem for anyone then tell me, and ill fix it)
*/
extern void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
/**
*
@@ -111,8 +110,8 @@ extern void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint
* problem for anyone then tell me, and ill fix it)
*/
extern void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride);
+ long width, long height,
+ long lumStride, long chromStride, long dstStride);
/**
*
@@ -121,26 +120,26 @@ extern void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_
* chrominance data is only taken from every secound line others are ignored FIXME write HQ version
*/
extern void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height,
- long lumStride, long chromStride, long srcStride);
+ long width, long height,
+ long lumStride, long chromStride, long srcStride);
extern void (*planar2x)(const uint8_t *src, uint8_t *dst, long width, long height,
- long srcStride, long dstStride);
+ long srcStride, long dstStride);
extern void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
- long width, long height, long src1Stride,
- long src2Stride, long dstStride);
+ long width, long height, long src1Stride,
+ long src2Stride, long dstStride);
extern void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
- uint8_t *dst1, uint8_t *dst2,
- long width, long height,
- long srcStride1, long srcStride2,
- long dstStride1, long dstStride2);
+ uint8_t *dst1, uint8_t *dst2,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long dstStride1, long dstStride2);
extern void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
- uint8_t *dst,
- long width, long height,
- long srcStride1, long srcStride2,
- long srcStride3, long dstStride);
+ uint8_t *dst,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long srcStride3, long dstStride);
void sws_rgb2rgb_init(int flags);
diff --git a/libswscale/rgb2rgb_template.c b/libswscale/rgb2rgb_template.c
index c089e39fa7..6dab0aa539 100644
--- a/libswscale/rgb2rgb_template.c
+++ b/libswscale/rgb2rgb_template.c
@@ -1,5 +1,4 @@
/*
- *
* rgb2rgb.c, Software RGB to RGB convertor
* pluralize by Software PAL8 to RGB convertor
* Software YUV to YUV convertor
@@ -53,11 +52,11 @@
#ifdef HAVE_3DNOW
#define PREFETCH "prefetch"
#define PREFETCHW "prefetchw"
-#define PAVGB "pavgusb"
+#define PAVGB "pavgusb"
#elif defined ( HAVE_MMX2 )
#define PREFETCH "prefetchnta"
#define PREFETCHW "prefetcht0"
-#define PAVGB "pavgb"
+#define PAVGB "pavgb"
#else
#ifdef __APPLE__
#define PREFETCH "#"
@@ -85,148 +84,148 @@
static inline void RENAME(rgb24to32)(const uint8_t *src,uint8_t *dst,long src_size)
{
- uint8_t *dest = dst;
- const uint8_t *s = src;
- const uint8_t *end;
-#ifdef HAVE_MMX
- const uint8_t *mm_end;
-#endif
- end = s + src_size;
-#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- mm_end = end - 23;
- __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
- while(s < mm_end)
- {
- __asm __volatile(
- PREFETCH" 32%1\n\t"
- "movd %1, %%mm0\n\t"
- "punpckldq 3%1, %%mm0\n\t"
- "movd 6%1, %%mm1\n\t"
- "punpckldq 9%1, %%mm1\n\t"
- "movd 12%1, %%mm2\n\t"
- "punpckldq 15%1, %%mm2\n\t"
- "movd 18%1, %%mm3\n\t"
- "punpckldq 21%1, %%mm3\n\t"
- "pand %%mm7, %%mm0\n\t"
- "pand %%mm7, %%mm1\n\t"
- "pand %%mm7, %%mm2\n\t"
- "pand %%mm7, %%mm3\n\t"
- MOVNTQ" %%mm0, %0\n\t"
- MOVNTQ" %%mm1, 8%0\n\t"
- MOVNTQ" %%mm2, 16%0\n\t"
- MOVNTQ" %%mm3, 24%0"
- :"=m"(*dest)
- :"m"(*s)
- :"memory");
- dest += 32;
- s += 24;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
-#ifdef WORDS_BIGENDIAN
- /* RGB24 (= R,G,B) -> RGB32 (= A,B,G,R) */
- *dest++ = 0;
- *dest++ = s[2];
- *dest++ = s[1];
- *dest++ = s[0];
- s+=3;
-#else
- *dest++ = *s++;
- *dest++ = *s++;
- *dest++ = *s++;
- *dest++ = 0;
-#endif
- }
+ uint8_t *dest = dst;
+ const uint8_t *s = src;
+ const uint8_t *end;
+ #ifdef HAVE_MMX
+ const uint8_t *mm_end;
+ #endif
+ end = s + src_size;
+ #ifdef HAVE_MMX
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ mm_end = end - 23;
+ __asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
+ while (s < mm_end)
+ {
+ __asm __volatile(
+ PREFETCH" 32%1 \n\t"
+ "movd %1, %%mm0 \n\t"
+ "punpckldq 3%1, %%mm0 \n\t"
+ "movd 6%1, %%mm1 \n\t"
+ "punpckldq 9%1, %%mm1 \n\t"
+ "movd 12%1, %%mm2 \n\t"
+ "punpckldq 15%1, %%mm2 \n\t"
+ "movd 18%1, %%mm3 \n\t"
+ "punpckldq 21%1, %%mm3 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm7, %%mm1 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "pand %%mm7, %%mm3 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm1, 8%0 \n\t"
+ MOVNTQ" %%mm2, 16%0 \n\t"
+ MOVNTQ" %%mm3, 24%0"
+ :"=m"(*dest)
+ :"m"(*s)
+ :"memory");
+ dest += 32;
+ s += 24;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+ #endif
+ while (s < end)
+ {
+ #ifdef WORDS_BIGENDIAN
+ /* RGB24 (= R,G,B) -> RGB32 (= A,B,G,R) */
+ *dest++ = 0;
+ *dest++ = s[2];
+ *dest++ = s[1];
+ *dest++ = s[0];
+ s+=3;
+ #else
+ *dest++ = *s++;
+ *dest++ = *s++;
+ *dest++ = *s++;
+ *dest++ = 0;
+ #endif
+ }
}
static inline void RENAME(rgb32to24)(const uint8_t *src,uint8_t *dst,long src_size)
{
- uint8_t *dest = dst;
- const uint8_t *s = src;
- const uint8_t *end;
+ uint8_t *dest = dst;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- end = s + src_size;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- mm_end = end - 31;
- 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");
- dest += 24;
- s += 32;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ mm_end = end - 31;
+ 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");
+ dest += 24;
+ s += 32;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
#ifdef WORDS_BIGENDIAN
- /* RGB32 (= A,B,G,R) -> RGB24 (= R,G,B) */
- s++;
- dest[2] = *s++;
- dest[1] = *s++;
- dest[0] = *s++;
- dest += 3;
+ /* RGB32 (= A,B,G,R) -> RGB24 (= R,G,B) */
+ s++;
+ dest[2] = *s++;
+ dest[1] = *s++;
+ dest[0] = *s++;
+ dest += 3;
#else
- *dest++ = *s++;
- *dest++ = *s++;
- *dest++ = *s++;
- s++;
+ *dest++ = *s++;
+ *dest++ = *s++;
+ *dest++ = *s++;
+ s++;
#endif
- }
+ }
}
/*
@@ -237,677 +236,677 @@ static inline void RENAME(rgb32to24)(const uint8_t *src,uint8_t *dst,long src_si
*/
static inline void RENAME(rgb15to16)(const uint8_t *src,uint8_t *dst,long src_size)
{
- register const uint8_t* s=src;
- register uint8_t* d=dst;
- register const uint8_t *end;
- const uint8_t *mm_end;
- end = s + src_size;
+ register const uint8_t* s=src;
+ register uint8_t* d=dst;
+ register const uint8_t *end;
+ const uint8_t *mm_end;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s));
- __asm __volatile("movq %0, %%mm4"::"m"(mask15s));
- mm_end = end - 15;
- 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)
- );
- d+=16;
- s+=16;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
+ __asm __volatile(PREFETCH" %0"::"m"(*s));
+ __asm __volatile("movq %0, %%mm4"::"m"(mask15s));
+ mm_end = end - 15;
+ 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)
+ );
+ d+=16;
+ s+=16;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
#endif
mm_end = end - 3;
- while(s < mm_end)
+ while (s < mm_end)
{
- register unsigned x= *((uint32_t *)s);
- *((uint32_t *)d) = (x&0x7FFF7FFF) + (x&0x7FE07FE0);
- d+=4;
- s+=4;
+ register unsigned x= *((uint32_t *)s);
+ *((uint32_t *)d) = (x&0x7FFF7FFF) + (x&0x7FE07FE0);
+ d+=4;
+ s+=4;
}
- if(s < end)
+ if (s < end)
{
- register unsigned short x= *((uint16_t *)s);
- *((uint16_t *)d) = (x&0x7FFF) + (x&0x7FE0);
+ register unsigned short x= *((uint16_t *)s);
+ *((uint16_t *)d) = (x&0x7FFF) + (x&0x7FE0);
}
}
static inline void RENAME(rgb16to15)(const uint8_t *src,uint8_t *dst,long src_size)
{
- register const uint8_t* s=src;
- register uint8_t* d=dst;
- register const uint8_t *end;
- const uint8_t *mm_end;
- end = s + src_size;
+ register const uint8_t* s=src;
+ register uint8_t* d=dst;
+ register const uint8_t *end;
+ const uint8_t *mm_end;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s));
- __asm __volatile("movq %0, %%mm7"::"m"(mask15rg));
- __asm __volatile("movq %0, %%mm6"::"m"(mask15b));
- mm_end = end - 15;
- 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)
- );
- d+=16;
- s+=16;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
+ __asm __volatile(PREFETCH" %0"::"m"(*s));
+ __asm __volatile("movq %0, %%mm7"::"m"(mask15rg));
+ __asm __volatile("movq %0, %%mm6"::"m"(mask15b));
+ mm_end = end - 15;
+ 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)
+ );
+ d+=16;
+ s+=16;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
#endif
mm_end = end - 3;
- while(s < mm_end)
+ while (s < mm_end)
{
- register uint32_t x= *((uint32_t *)s);
- *((uint32_t *)d) = ((x>>1)&0x7FE07FE0) | (x&0x001F001F);
- s+=4;
- d+=4;
+ register uint32_t x= *((uint32_t *)s);
+ *((uint32_t *)d) = ((x>>1)&0x7FE07FE0) | (x&0x001F001F);
+ s+=4;
+ d+=4;
}
- if(s < end)
+ if (s < end)
{
- register uint16_t x= *((uint16_t *)s);
- *((uint16_t *)d) = ((x>>1)&0x7FE0) | (x&0x001F);
- s+=2;
- d+=2;
+ register uint16_t x= *((uint16_t *)s);
+ *((uint16_t *)d) = ((x>>1)&0x7FE0) | (x&0x001F);
+ s+=2;
+ d+=2;
}
}
static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- mm_end = end - 15;
+ mm_end = end - 15;
#if 1 //is faster only if multiplies are reasonable fast (FIXME figure out on which cpus this is faster, on Athlon its 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)
- );
+ 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)
+ );
#else
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_16mask),"m"(green_16mask));
- 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");
- d += 4;
- s += 16;
- }
-#endif
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register int rgb = *(uint32_t*)s; s += 4;
- *d++ = ((rgb&0xFF)>>3) + ((rgb&0xFC00)>>5) + ((rgb&0xF80000)>>8);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_16mask),"m"(green_16mask));
+ 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");
+ d += 4;
+ s += 16;
+ }
+#endif
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register int rgb = *(uint32_t*)s; s += 4;
+ *d++ = ((rgb&0xFF)>>3) + ((rgb&0xFC00)>>5) + ((rgb&0xF80000)>>8);
+ }
}
static inline void RENAME(rgb32tobgr16)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_16mask),"m"(green_16mask));
- mm_end = end - 15;
- 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");
- d += 4;
- s += 16;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register int rgb = *(uint32_t*)s; s += 4;
- *d++ = ((rgb&0xF8)<<8) + ((rgb&0xFC00)>>5) + ((rgb&0xF80000)>>19);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_16mask),"m"(green_16mask));
+ mm_end = end - 15;
+ 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");
+ d += 4;
+ s += 16;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register int rgb = *(uint32_t*)s; s += 4;
+ *d++ = ((rgb&0xF8)<<8) + ((rgb&0xFC00)>>5) + ((rgb&0xF80000)>>19);
+ }
}
static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- mm_end = end - 15;
+ mm_end = end - 15;
#if 1 //is faster only if multiplies are reasonable fast (FIXME figure out on which cpus this is faster, on Athlon its 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)
- );
+ 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)
+ );
#else
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_15mask),"m"(green_15mask));
- 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");
- d += 4;
- s += 16;
- }
-#endif
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register int rgb = *(uint32_t*)s; s += 4;
- *d++ = ((rgb&0xFF)>>3) + ((rgb&0xF800)>>6) + ((rgb&0xF80000)>>9);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_15mask),"m"(green_15mask));
+ 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");
+ d += 4;
+ s += 16;
+ }
+#endif
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register int rgb = *(uint32_t*)s; s += 4;
+ *d++ = ((rgb&0xFF)>>3) + ((rgb&0xF800)>>6) + ((rgb&0xF80000)>>9);
+ }
}
static inline void RENAME(rgb32tobgr15)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_15mask),"m"(green_15mask));
- mm_end = end - 15;
- 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");
- d += 4;
- s += 16;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register int rgb = *(uint32_t*)s; s += 4;
- *d++ = ((rgb&0xF8)<<7) + ((rgb&0xF800)>>6) + ((rgb&0xF80000)>>19);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_15mask),"m"(green_15mask));
+ mm_end = end - 15;
+ 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");
+ d += 4;
+ s += 16;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register int rgb = *(uint32_t*)s; s += 4;
+ *d++ = ((rgb&0xF8)<<7) + ((rgb&0xF800)>>6) + ((rgb&0xF80000)>>19);
+ }
}
static inline void RENAME(rgb24to16)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_16mask),"m"(green_16mask));
- mm_end = end - 11;
- 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");
- d += 4;
- s += 12;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- const int b= *s++;
- const int g= *s++;
- const int r= *s++;
- *d++ = (b>>3) | ((g&0xFC)<<3) | ((r&0xF8)<<8);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_16mask),"m"(green_16mask));
+ mm_end = end - 11;
+ 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");
+ d += 4;
+ s += 12;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ const int b = *s++;
+ const int g = *s++;
+ const int r = *s++;
+ *d++ = (b>>3) | ((g&0xFC)<<3) | ((r&0xF8)<<8);
+ }
}
static inline void RENAME(rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_16mask),"m"(green_16mask));
- mm_end = end - 15;
- 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"
- "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 += 12;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- const int r= *s++;
- const int g= *s++;
- const int b= *s++;
- *d++ = (b>>3) | ((g&0xFC)<<3) | ((r&0xF8)<<8);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_16mask),"m"(green_16mask));
+ mm_end = end - 15;
+ 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"
+ "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 += 12;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ const int r = *s++;
+ const int g = *s++;
+ const int b = *s++;
+ *d++ = (b>>3) | ((g&0xFC)<<3) | ((r&0xF8)<<8);
+ }
}
static inline void RENAME(rgb24to15)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_15mask),"m"(green_15mask));
- mm_end = end - 11;
- 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 $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 += 12;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- const int b= *s++;
- const int g= *s++;
- const int r= *s++;
- *d++ = (b>>3) | ((g&0xF8)<<2) | ((r&0xF8)<<7);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_15mask),"m"(green_15mask));
+ mm_end = end - 11;
+ 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 $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 += 12;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ const int b = *s++;
+ const int g = *s++;
+ const int r = *s++;
+ *d++ = (b>>3) | ((g&0xF8)<<2) | ((r&0xF8)<<7);
+ }
}
static inline void RENAME(rgb24tobgr15)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint8_t *s = src;
- const uint8_t *end;
+ const uint8_t *s = src;
+ const uint8_t *end;
#ifdef HAVE_MMX
- const uint8_t *mm_end;
+ const uint8_t *mm_end;
#endif
- uint16_t *d = (uint16_t *)dst;
- end = s + src_size;
+ uint16_t *d = (uint16_t *)dst;
+ end = s + src_size;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
- __asm __volatile(
- "movq %0, %%mm7\n\t"
- "movq %1, %%mm6\n\t"
- ::"m"(red_15mask),"m"(green_15mask));
- mm_end = end - 15;
- 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"
- "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 += 12;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- const int r= *s++;
- const int g= *s++;
- const int b= *s++;
- *d++ = (b>>3) | ((g&0xF8)<<2) | ((r&0xF8)<<7);
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*src):"memory");
+ __asm __volatile(
+ "movq %0, %%mm7 \n\t"
+ "movq %1, %%mm6 \n\t"
+ ::"m"(red_15mask),"m"(green_15mask));
+ mm_end = end - 15;
+ 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"
+ "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 += 12;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ const int r = *s++;
+ const int g = *s++;
+ const int b = *s++;
+ *d++ = (b>>3) | ((g&0xF8)<<2) | ((r&0xF8)<<7);
+ }
}
/*
@@ -935,706 +934,706 @@ static inline void RENAME(rgb24tobgr15)(const uint8_t *src, uint8_t *dst, long s
*/
static inline void RENAME(rgb15to24)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
+ const uint16_t *end;
#ifdef HAVE_MMX
- const uint16_t *mm_end;
+ const uint16_t *mm_end;
#endif
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (uint16_t *)src;
- end = s + src_size/2;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (uint16_t *)src;
+ end = s + src_size/2;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- mm_end = end - 7;
- while(s < mm_end)
- {
- __asm __volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq %1, %%mm1\n\t"
- "movq %1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $2, %%mm1\n\t"
- "psrlq $7, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %5, %%mm0\n\t"
- "punpcklwd %5, %%mm1\n\t"
- "punpcklwd %5, %%mm2\n\t"
- "punpckhwd %5, %%mm3\n\t"
- "punpckhwd %5, %%mm4\n\t"
- "punpckhwd %5, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
-
- "movq %%mm0, %%mm6\n\t"
- "movq %%mm3, %%mm7\n\t"
-
- "movq 8%1, %%mm0\n\t"
- "movq 8%1, %%mm1\n\t"
- "movq 8%1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $2, %%mm1\n\t"
- "psrlq $7, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %5, %%mm0\n\t"
- "punpcklwd %5, %%mm1\n\t"
- "punpcklwd %5, %%mm2\n\t"
- "punpckhwd %5, %%mm3\n\t"
- "punpckhwd %5, %%mm4\n\t"
- "punpckhwd %5, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
-
- :"=m"(*d)
- :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r), "m"(mmx_null)
- :"memory");
- /* Borrowed 32 to 24 */
- __asm __volatile(
- "movq %%mm0, %%mm4\n\t"
- "movq %%mm3, %%mm5\n\t"
- "movq %%mm6, %%mm0\n\t"
- "movq %%mm7, %%mm1\n\t"
-
- "movq %%mm4, %%mm6\n\t"
- "movq %%mm5, %%mm7\n\t"
- "movq %%mm0, %%mm2\n\t"
- "movq %%mm1, %%mm3\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"(*d)
- :"m"(*s),"m"(mask24l),"m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
- :"memory");
- d += 24;
- s += 8;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x7C00)>>7;
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ mm_end = end - 7;
+ while (s < mm_end)
+ {
+ __asm __volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq %1, %%mm1 \n\t"
+ "movq %1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $2, %%mm1 \n\t"
+ "psrlq $7, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %5, %%mm0 \n\t"
+ "punpcklwd %5, %%mm1 \n\t"
+ "punpcklwd %5, %%mm2 \n\t"
+ "punpckhwd %5, %%mm3 \n\t"
+ "punpckhwd %5, %%mm4 \n\t"
+ "punpckhwd %5, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+
+ "movq %%mm0, %%mm6 \n\t"
+ "movq %%mm3, %%mm7 \n\t"
+
+ "movq 8%1, %%mm0 \n\t"
+ "movq 8%1, %%mm1 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $2, %%mm1 \n\t"
+ "psrlq $7, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %5, %%mm0 \n\t"
+ "punpcklwd %5, %%mm1 \n\t"
+ "punpcklwd %5, %%mm2 \n\t"
+ "punpckhwd %5, %%mm3 \n\t"
+ "punpckhwd %5, %%mm4 \n\t"
+ "punpckhwd %5, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+
+ :"=m"(*d)
+ :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r), "m"(mmx_null)
+ :"memory");
+ /* Borrowed 32 to 24 */
+ __asm __volatile(
+ "movq %%mm0, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "movq %%mm6, %%mm0 \n\t"
+ "movq %%mm7, %%mm1 \n\t"
+
+ "movq %%mm4, %%mm6 \n\t"
+ "movq %%mm5, %%mm7 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm1, %%mm3 \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"(*d)
+ :"m"(*s),"m"(mask24l),"m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
+ :"memory");
+ d += 24;
+ s += 8;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x7C00)>>7;
+ }
}
static inline void RENAME(rgb16to24)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
+ const uint16_t *end;
#ifdef HAVE_MMX
- const uint16_t *mm_end;
+ const uint16_t *mm_end;
#endif
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (const uint16_t *)src;
- end = s + src_size/2;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (const uint16_t *)src;
+ end = s + src_size/2;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- mm_end = end - 7;
- while(s < mm_end)
- {
- __asm __volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq %1, %%mm1\n\t"
- "movq %1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $3, %%mm1\n\t"
- "psrlq $8, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %5, %%mm0\n\t"
- "punpcklwd %5, %%mm1\n\t"
- "punpcklwd %5, %%mm2\n\t"
- "punpckhwd %5, %%mm3\n\t"
- "punpckhwd %5, %%mm4\n\t"
- "punpckhwd %5, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
-
- "movq %%mm0, %%mm6\n\t"
- "movq %%mm3, %%mm7\n\t"
-
- "movq 8%1, %%mm0\n\t"
- "movq 8%1, %%mm1\n\t"
- "movq 8%1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $3, %%mm1\n\t"
- "psrlq $8, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %5, %%mm0\n\t"
- "punpcklwd %5, %%mm1\n\t"
- "punpcklwd %5, %%mm2\n\t"
- "punpckhwd %5, %%mm3\n\t"
- "punpckhwd %5, %%mm4\n\t"
- "punpckhwd %5, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
- :"=m"(*d)
- :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mmx_null)
- :"memory");
- /* Borrowed 32 to 24 */
- __asm __volatile(
- "movq %%mm0, %%mm4\n\t"
- "movq %%mm3, %%mm5\n\t"
- "movq %%mm6, %%mm0\n\t"
- "movq %%mm7, %%mm1\n\t"
-
- "movq %%mm4, %%mm6\n\t"
- "movq %%mm5, %%mm7\n\t"
- "movq %%mm0, %%mm2\n\t"
- "movq %%mm1, %%mm3\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"(*d)
- :"m"(*s),"m"(mask24l),"m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
- :"memory");
- d += 24;
- s += 8;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0xF800)>>8;
- }
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ mm_end = end - 7;
+ while (s < mm_end)
+ {
+ __asm __volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq %1, %%mm1 \n\t"
+ "movq %1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm1 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %5, %%mm0 \n\t"
+ "punpcklwd %5, %%mm1 \n\t"
+ "punpcklwd %5, %%mm2 \n\t"
+ "punpckhwd %5, %%mm3 \n\t"
+ "punpckhwd %5, %%mm4 \n\t"
+ "punpckhwd %5, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+
+ "movq %%mm0, %%mm6 \n\t"
+ "movq %%mm3, %%mm7 \n\t"
+
+ "movq 8%1, %%mm0 \n\t"
+ "movq 8%1, %%mm1 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm1 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %5, %%mm0 \n\t"
+ "punpcklwd %5, %%mm1 \n\t"
+ "punpcklwd %5, %%mm2 \n\t"
+ "punpckhwd %5, %%mm3 \n\t"
+ "punpckhwd %5, %%mm4 \n\t"
+ "punpckhwd %5, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ :"=m"(*d)
+ :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mmx_null)
+ :"memory");
+ /* Borrowed 32 to 24 */
+ __asm __volatile(
+ "movq %%mm0, %%mm4 \n\t"
+ "movq %%mm3, %%mm5 \n\t"
+ "movq %%mm6, %%mm0 \n\t"
+ "movq %%mm7, %%mm1 \n\t"
+
+ "movq %%mm4, %%mm6 \n\t"
+ "movq %%mm5, %%mm7 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm1, %%mm3 \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"(*d)
+ :"m"(*s),"m"(mask24l),"m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
+ :"memory");
+ d += 24;
+ s += 8;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0xF800)>>8;
+ }
}
static inline void RENAME(rgb15to32)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
+ const uint16_t *end;
#ifdef HAVE_MMX
- const uint16_t *mm_end;
+ const uint16_t *mm_end;
#endif
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (const uint16_t *)src;
- end = s + src_size/2;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (const uint16_t *)src;
+ end = s + src_size/2;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- __asm __volatile("pxor %%mm7,%%mm7\n\t":::"memory");
- mm_end = end - 3;
- while(s < mm_end)
- {
- __asm __volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq %1, %%mm1\n\t"
- "movq %1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $2, %%mm1\n\t"
- "psrlq $7, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %%mm7, %%mm0\n\t"
- "punpcklwd %%mm7, %%mm1\n\t"
- "punpcklwd %%mm7, %%mm2\n\t"
- "punpckhwd %%mm7, %%mm3\n\t"
- "punpckhwd %%mm7, %%mm4\n\t"
- "punpckhwd %%mm7, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
- MOVNTQ" %%mm0, %0\n\t"
- MOVNTQ" %%mm3, 8%0\n\t"
- :"=m"(*d)
- :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r)
- :"memory");
- d += 16;
- s += 4;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ __asm __volatile("pxor %%mm7,%%mm7 \n\t":::"memory");
+ mm_end = end - 3;
+ while (s < mm_end)
+ {
+ __asm __volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq %1, %%mm1 \n\t"
+ "movq %1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $2, %%mm1 \n\t"
+ "psrlq $7, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %%mm7, %%mm0 \n\t"
+ "punpcklwd %%mm7, %%mm1 \n\t"
+ "punpcklwd %%mm7, %%mm2 \n\t"
+ "punpckhwd %%mm7, %%mm3 \n\t"
+ "punpckhwd %%mm7, %%mm4 \n\t"
+ "punpckhwd %%mm7, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm3, 8%0 \n\t"
+ :"=m"(*d)
+ :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r)
+ :"memory");
+ d += 16;
+ s += 4;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
#if 0 //slightly slower on athlon
- int bgr= *s++;
- *((uint32_t*)d)++ = ((bgr&0x1F)<<3) + ((bgr&0x3E0)<<6) + ((bgr&0x7C00)<<9);
+ int bgr= *s++;
+ *((uint32_t*)d)++ = ((bgr&0x1F)<<3) + ((bgr&0x3E0)<<6) + ((bgr&0x7C00)<<9);
#else
- register uint16_t bgr;
- bgr = *s++;
+ register uint16_t bgr;
+ bgr = *s++;
#ifdef WORDS_BIGENDIAN
- *d++ = 0;
- *d++ = (bgr&0x7C00)>>7;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x1F)<<3;
+ *d++ = 0;
+ *d++ = (bgr&0x7C00)>>7;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x1F)<<3;
#else
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x3E0)>>2;
- *d++ = (bgr&0x7C00)>>7;
- *d++ = 0;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x3E0)>>2;
+ *d++ = (bgr&0x7C00)>>7;
+ *d++ = 0;
#endif
#endif
- }
+ }
}
static inline void RENAME(rgb16to32)(const uint8_t *src, uint8_t *dst, long src_size)
{
- const uint16_t *end;
+ const uint16_t *end;
#ifdef HAVE_MMX
- const uint16_t *mm_end;
+ const uint16_t *mm_end;
#endif
- uint8_t *d = (uint8_t *)dst;
- const uint16_t *s = (uint16_t *)src;
- end = s + src_size/2;
+ uint8_t *d = (uint8_t *)dst;
+ const uint16_t *s = (uint16_t *)src;
+ end = s + src_size/2;
#ifdef HAVE_MMX
- __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
- __asm __volatile("pxor %%mm7,%%mm7\n\t":::"memory");
- mm_end = end - 3;
- while(s < mm_end)
- {
- __asm __volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq %1, %%mm1\n\t"
- "movq %1, %%mm2\n\t"
- "pand %2, %%mm0\n\t"
- "pand %3, %%mm1\n\t"
- "pand %4, %%mm2\n\t"
- "psllq $3, %%mm0\n\t"
- "psrlq $3, %%mm1\n\t"
- "psrlq $8, %%mm2\n\t"
- "movq %%mm0, %%mm3\n\t"
- "movq %%mm1, %%mm4\n\t"
- "movq %%mm2, %%mm5\n\t"
- "punpcklwd %%mm7, %%mm0\n\t"
- "punpcklwd %%mm7, %%mm1\n\t"
- "punpcklwd %%mm7, %%mm2\n\t"
- "punpckhwd %%mm7, %%mm3\n\t"
- "punpckhwd %%mm7, %%mm4\n\t"
- "punpckhwd %%mm7, %%mm5\n\t"
- "psllq $8, %%mm1\n\t"
- "psllq $16, %%mm2\n\t"
- "por %%mm1, %%mm0\n\t"
- "por %%mm2, %%mm0\n\t"
- "psllq $8, %%mm4\n\t"
- "psllq $16, %%mm5\n\t"
- "por %%mm4, %%mm3\n\t"
- "por %%mm5, %%mm3\n\t"
- MOVNTQ" %%mm0, %0\n\t"
- MOVNTQ" %%mm3, 8%0\n\t"
- :"=m"(*d)
- :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r)
- :"memory");
- d += 16;
- s += 4;
- }
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-#endif
- while(s < end)
- {
- register uint16_t bgr;
- bgr = *s++;
+ __asm __volatile(PREFETCH" %0"::"m"(*s):"memory");
+ __asm __volatile("pxor %%mm7,%%mm7 \n\t":::"memory");
+ mm_end = end - 3;
+ while (s < mm_end)
+ {
+ __asm __volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq %1, %%mm1 \n\t"
+ "movq %1, %%mm2 \n\t"
+ "pand %2, %%mm0 \n\t"
+ "pand %3, %%mm1 \n\t"
+ "pand %4, %%mm2 \n\t"
+ "psllq $3, %%mm0 \n\t"
+ "psrlq $3, %%mm1 \n\t"
+ "psrlq $8, %%mm2 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "movq %%mm2, %%mm5 \n\t"
+ "punpcklwd %%mm7, %%mm0 \n\t"
+ "punpcklwd %%mm7, %%mm1 \n\t"
+ "punpcklwd %%mm7, %%mm2 \n\t"
+ "punpckhwd %%mm7, %%mm3 \n\t"
+ "punpckhwd %%mm7, %%mm4 \n\t"
+ "punpckhwd %%mm7, %%mm5 \n\t"
+ "psllq $8, %%mm1 \n\t"
+ "psllq $16, %%mm2 \n\t"
+ "por %%mm1, %%mm0 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "psllq $8, %%mm4 \n\t"
+ "psllq $16, %%mm5 \n\t"
+ "por %%mm4, %%mm3 \n\t"
+ "por %%mm5, %%mm3 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm3, 8%0 \n\t"
+ :"=m"(*d)
+ :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r)
+ :"memory");
+ d += 16;
+ s += 4;
+ }
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+#endif
+ while (s < end)
+ {
+ register uint16_t bgr;
+ bgr = *s++;
#ifdef WORDS_BIGENDIAN
- *d++ = 0;
- *d++ = (bgr&0xF800)>>8;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0x1F)<<3;
+ *d++ = 0;
+ *d++ = (bgr&0xF800)>>8;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0x1F)<<3;
#else
- *d++ = (bgr&0x1F)<<3;
- *d++ = (bgr&0x7E0)>>3;
- *d++ = (bgr&0xF800)>>8;
- *d++ = 0;
+ *d++ = (bgr&0x1F)<<3;
+ *d++ = (bgr&0x7E0)>>3;
+ *d++ = (bgr&0xF800)>>8;
+ *d++ = 0;
#endif
- }
+ }
}
static inline void RENAME(rgb32tobgr32)(const uint8_t *src, uint8_t *dst, long src_size)
{
- long idx = 15 - src_size;
- uint8_t *s = (uint8_t *) src-idx, *d = dst-idx;
+ long idx = 15 - src_size;
+ uint8_t *s = (uint8_t *) src-idx, *d = dst-idx;
#ifdef HAVE_MMX
- __asm __volatile(
- "test %0, %0 \n\t"
- "jns 2f \n\t"
- PREFETCH" (%1, %0) \n\t"
- "movq %3, %%mm7 \n\t"
- "pxor %4, %%mm7 \n\t"
- "movq %%mm7, %%mm6 \n\t"
- "pxor %5, %%mm7 \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1, %0) \n\t"
- "movq (%1, %0), %%mm0 \n\t"
- "movq 8(%1, %0), %%mm1 \n\t"
+ __asm __volatile(
+ "test %0, %0 \n\t"
+ "jns 2f \n\t"
+ PREFETCH" (%1, %0) \n\t"
+ "movq %3, %%mm7 \n\t"
+ "pxor %4, %%mm7 \n\t"
+ "movq %%mm7, %%mm6 \n\t"
+ "pxor %5, %%mm7 \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1, %0) \n\t"
+ "movq (%1, %0), %%mm0 \n\t"
+ "movq 8(%1, %0), %%mm1 \n\t"
# ifdef HAVE_MMX2
- "pshufw $177, %%mm0, %%mm3 \n\t"
- "pshufw $177, %%mm1, %%mm5 \n\t"
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm6, %%mm3 \n\t"
- "pand %%mm7, %%mm1 \n\t"
- "pand %%mm6, %%mm5 \n\t"
- "por %%mm3, %%mm0 \n\t"
- "por %%mm5, %%mm1 \n\t"
+ "pshufw $177, %%mm0, %%mm3 \n\t"
+ "pshufw $177, %%mm1, %%mm5 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm6, %%mm3 \n\t"
+ "pand %%mm7, %%mm1 \n\t"
+ "pand %%mm6, %%mm5 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ "por %%mm5, %%mm1 \n\t"
# else
- "movq %%mm0, %%mm2 \n\t"
- "movq %%mm1, %%mm4 \n\t"
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm6, %%mm2 \n\t"
- "pand %%mm7, %%mm1 \n\t"
- "pand %%mm6, %%mm4 \n\t"
- "movq %%mm2, %%mm3 \n\t"
- "movq %%mm4, %%mm5 \n\t"
- "pslld $16, %%mm2 \n\t"
- "psrld $16, %%mm3 \n\t"
- "pslld $16, %%mm4 \n\t"
- "psrld $16, %%mm5 \n\t"
- "por %%mm2, %%mm0 \n\t"
- "por %%mm4, %%mm1 \n\t"
- "por %%mm3, %%mm0 \n\t"
- "por %%mm5, %%mm1 \n\t"
+ "movq %%mm0, %%mm2 \n\t"
+ "movq %%mm1, %%mm4 \n\t"
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm6, %%mm2 \n\t"
+ "pand %%mm7, %%mm1 \n\t"
+ "pand %%mm6, %%mm4 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "movq %%mm4, %%mm5 \n\t"
+ "pslld $16, %%mm2 \n\t"
+ "psrld $16, %%mm3 \n\t"
+ "pslld $16, %%mm4 \n\t"
+ "psrld $16, %%mm5 \n\t"
+ "por %%mm2, %%mm0 \n\t"
+ "por %%mm4, %%mm1 \n\t"
+ "por %%mm3, %%mm0 \n\t"
+ "por %%mm5, %%mm1 \n\t"
# endif
- MOVNTQ" %%mm0, (%2, %0) \n\t"
- MOVNTQ" %%mm1, 8(%2, %0) \n\t"
- "add $16, %0 \n\t"
- "js 1b \n\t"
- SFENCE" \n\t"
- EMMS" \n\t"
- "2: \n\t"
- : "+&r"(idx)
- : "r" (s), "r" (d), "m" (mask32b), "m" (mask32r), "m" (mmx_one)
- : "memory");
-#endif
- for (; idx<15; idx+=4) {
- register int v = *(uint32_t *)&s[idx], g = v & 0xff00ff00;
- v &= 0xff00ff;
- *(uint32_t *)&d[idx] = (v>>16) + g + (v<<16);
- }
+ MOVNTQ" %%mm0, (%2, %0) \n\t"
+ MOVNTQ" %%mm1, 8(%2, %0) \n\t"
+ "add $16, %0 \n\t"
+ "js 1b \n\t"
+ SFENCE" \n\t"
+ EMMS" \n\t"
+ "2: \n\t"
+ : "+&r"(idx)
+ : "r" (s), "r" (d), "m" (mask32b), "m" (mask32r), "m" (mmx_one)
+ : "memory");
+#endif
+ for (; idx<15; idx+=4) {
+ register int v = *(uint32_t *)&s[idx], g = v & 0xff00ff00;
+ v &= 0xff00ff;
+ *(uint32_t *)&d[idx] = (v>>16) + g + (v<<16);
+ }
}
static inline void RENAME(rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long src_size)
{
- unsigned i;
+ unsigned i;
#ifdef HAVE_MMX
- long mmx_size= 23 - src_size;
- asm volatile (
- "test %%"REG_a", %%"REG_a" \n\t"
- "jns 2f \n\t"
- "movq "MANGLE(mask24r)", %%mm5 \n\t"
- "movq "MANGLE(mask24g)", %%mm6 \n\t"
- "movq "MANGLE(mask24b)", %%mm7 \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1, %%"REG_a") \n\t"
- "movq (%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG
- "movq (%1, %%"REG_a"), %%mm1 \n\t" // BGR BGR BG
- "movq 2(%1, %%"REG_a"), %%mm2 \n\t" // R BGR BGR B
- "psllq $16, %%mm0 \n\t" // 00 BGR BGR
- "pand %%mm5, %%mm0 \n\t"
- "pand %%mm6, %%mm1 \n\t"
- "pand %%mm7, %%mm2 \n\t"
- "por %%mm0, %%mm1 \n\t"
- "por %%mm2, %%mm1 \n\t"
- "movq 6(%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG
- MOVNTQ" %%mm1, (%2, %%"REG_a")\n\t" // RGB RGB RG
- "movq 8(%1, %%"REG_a"), %%mm1 \n\t" // R BGR BGR B
- "movq 10(%1, %%"REG_a"), %%mm2 \n\t" // GR BGR BGR
- "pand %%mm7, %%mm0 \n\t"
- "pand %%mm5, %%mm1 \n\t"
- "pand %%mm6, %%mm2 \n\t"
- "por %%mm0, %%mm1 \n\t"
- "por %%mm2, %%mm1 \n\t"
- "movq 14(%1, %%"REG_a"), %%mm0 \n\t" // R BGR BGR B
- MOVNTQ" %%mm1, 8(%2, %%"REG_a")\n\t" // B RGB RGB R
- "movq 16(%1, %%"REG_a"), %%mm1 \n\t" // GR BGR BGR
- "movq 18(%1, %%"REG_a"), %%mm2 \n\t" // BGR BGR BG
- "pand %%mm6, %%mm0 \n\t"
- "pand %%mm7, %%mm1 \n\t"
- "pand %%mm5, %%mm2 \n\t"
- "por %%mm0, %%mm1 \n\t"
- "por %%mm2, %%mm1 \n\t"
- MOVNTQ" %%mm1, 16(%2, %%"REG_a")\n\t"
- "add $24, %%"REG_a" \n\t"
- " js 1b \n\t"
- "2: \n\t"
- : "+a" (mmx_size)
- : "r" (src-mmx_size), "r"(dst-mmx_size)
- );
-
- __asm __volatile(SFENCE:::"memory");
- __asm __volatile(EMMS:::"memory");
-
- if(mmx_size==23) return; //finihsed, was multiple of 8
-
- src+= src_size;
- dst+= src_size;
- src_size= 23-mmx_size;
- src-= src_size;
- dst-= src_size;
-#endif
- for(i=0; i<src_size; i+=3)
- {
- register uint8_t x;
- x = src[i + 2];
- dst[i + 1] = src[i + 1];
- dst[i + 2] = src[i + 0];
- dst[i + 0] = x;
- }
+ long mmx_size= 23 - src_size;
+ asm volatile (
+ "test %%"REG_a", %%"REG_a" \n\t"
+ "jns 2f \n\t"
+ "movq "MANGLE(mask24r)", %%mm5 \n\t"
+ "movq "MANGLE(mask24g)", %%mm6 \n\t"
+ "movq "MANGLE(mask24b)", %%mm7 \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1, %%"REG_a") \n\t"
+ "movq (%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG
+ "movq (%1, %%"REG_a"), %%mm1 \n\t" // BGR BGR BG
+ "movq 2(%1, %%"REG_a"), %%mm2 \n\t" // R BGR BGR B
+ "psllq $16, %%mm0 \n\t" // 00 BGR BGR
+ "pand %%mm5, %%mm0 \n\t"
+ "pand %%mm6, %%mm1 \n\t"
+ "pand %%mm7, %%mm2 \n\t"
+ "por %%mm0, %%mm1 \n\t"
+ "por %%mm2, %%mm1 \n\t"
+ "movq 6(%1, %%"REG_a"), %%mm0 \n\t" // BGR BGR BG
+ MOVNTQ" %%mm1, (%2, %%"REG_a") \n\t" // RGB RGB RG
+ "movq 8(%1, %%"REG_a"), %%mm1 \n\t" // R BGR BGR B
+ "movq 10(%1, %%"REG_a"), %%mm2 \n\t" // GR BGR BGR
+ "pand %%mm7, %%mm0 \n\t"
+ "pand %%mm5, %%mm1 \n\t"
+ "pand %%mm6, %%mm2 \n\t"
+ "por %%mm0, %%mm1 \n\t"
+ "por %%mm2, %%mm1 \n\t"
+ "movq 14(%1, %%"REG_a"), %%mm0 \n\t" // R BGR BGR B
+ MOVNTQ" %%mm1, 8(%2, %%"REG_a") \n\t" // B RGB RGB R
+ "movq 16(%1, %%"REG_a"), %%mm1 \n\t" // GR BGR BGR
+ "movq 18(%1, %%"REG_a"), %%mm2 \n\t" // BGR BGR BG
+ "pand %%mm6, %%mm0 \n\t"
+ "pand %%mm7, %%mm1 \n\t"
+ "pand %%mm5, %%mm2 \n\t"
+ "por %%mm0, %%mm1 \n\t"
+ "por %%mm2, %%mm1 \n\t"
+ MOVNTQ" %%mm1, 16(%2, %%"REG_a") \n\t"
+ "add $24, %%"REG_a" \n\t"
+ " js 1b \n\t"
+ "2: \n\t"
+ : "+a" (mmx_size)
+ : "r" (src-mmx_size), "r"(dst-mmx_size)
+ );
+
+ __asm __volatile(SFENCE:::"memory");
+ __asm __volatile(EMMS:::"memory");
+
+ if (mmx_size==23) return; //finihsed, was multiple of 8
+
+ src+= src_size;
+ dst+= src_size;
+ src_size= 23-mmx_size;
+ src-= src_size;
+ dst-= src_size;
+#endif
+ for (i=0; i<src_size; i+=3)
+ {
+ register uint8_t x;
+ x = src[i + 2];
+ dst[i + 1] = src[i + 1];
+ dst[i + 2] = src[i + 0];
+ dst[i + 0] = x;
+ }
}
static inline void RENAME(yuvPlanartoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride, long vertLumPerChroma)
+ long width, long height,
+ long lumStride, long chromStride, long dstStride, long vertLumPerChroma)
{
- long y;
- const long chromWidth= width>>1;
- for(y=0; y<height; y++)
- {
+ long y;
+ const long chromWidth= width>>1;
+ for (y=0; y<height; y++)
+ {
#ifdef HAVE_MMX
//FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway)
- asm volatile(
- "xor %%"REG_a", %%"REG_a" \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
- PREFETCH" 32(%2, %%"REG_a") \n\t"
- PREFETCH" 32(%3, %%"REG_a") \n\t"
- "movq (%2, %%"REG_a"), %%mm0 \n\t" // U(0)
- "movq %%mm0, %%mm2 \n\t" // U(0)
- "movq (%3, %%"REG_a"), %%mm1 \n\t" // V(0)
- "punpcklbw %%mm1, %%mm0 \n\t" // UVUV UVUV(0)
- "punpckhbw %%mm1, %%mm2 \n\t" // UVUV UVUV(8)
-
- "movq (%1, %%"REG_a",2), %%mm3 \n\t" // Y(0)
- "movq 8(%1, %%"REG_a",2), %%mm5 \n\t" // Y(8)
- "movq %%mm3, %%mm4 \n\t" // Y(0)
- "movq %%mm5, %%mm6 \n\t" // Y(8)
- "punpcklbw %%mm0, %%mm3 \n\t" // YUYV YUYV(0)
- "punpckhbw %%mm0, %%mm4 \n\t" // YUYV YUYV(4)
- "punpcklbw %%mm2, %%mm5 \n\t" // YUYV YUYV(8)
- "punpckhbw %%mm2, %%mm6 \n\t" // YUYV YUYV(12)
-
- MOVNTQ" %%mm3, (%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm4, 8(%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm5, 16(%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm6, 24(%0, %%"REG_a", 4)\n\t"
-
- "add $8, %%"REG_a" \n\t"
- "cmp %4, %%"REG_a" \n\t"
- " jb 1b \n\t"
- ::"r"(dst), "r"(ysrc), "r"(usrc), "r"(vsrc), "g" (chromWidth)
- : "%"REG_a
- );
+ asm volatile(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
+ PREFETCH" 32(%2, %%"REG_a") \n\t"
+ PREFETCH" 32(%3, %%"REG_a") \n\t"
+ "movq (%2, %%"REG_a"), %%mm0 \n\t" // U(0)
+ "movq %%mm0, %%mm2 \n\t" // U(0)
+ "movq (%3, %%"REG_a"), %%mm1 \n\t" // V(0)
+ "punpcklbw %%mm1, %%mm0 \n\t" // UVUV UVUV(0)
+ "punpckhbw %%mm1, %%mm2 \n\t" // UVUV UVUV(8)
+
+ "movq (%1, %%"REG_a",2), %%mm3 \n\t" // Y(0)
+ "movq 8(%1, %%"REG_a",2), %%mm5 \n\t" // Y(8)
+ "movq %%mm3, %%mm4 \n\t" // Y(0)
+ "movq %%mm5, %%mm6 \n\t" // Y(8)
+ "punpcklbw %%mm0, %%mm3 \n\t" // YUYV YUYV(0)
+ "punpckhbw %%mm0, %%mm4 \n\t" // YUYV YUYV(4)
+ "punpcklbw %%mm2, %%mm5 \n\t" // YUYV YUYV(8)
+ "punpckhbw %%mm2, %%mm6 \n\t" // YUYV YUYV(12)
+
+ MOVNTQ" %%mm3, (%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm4, 8(%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm5, 16(%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm6, 24(%0, %%"REG_a", 4) \n\t"
+
+ "add $8, %%"REG_a" \n\t"
+ "cmp %4, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+ ::"r"(dst), "r"(ysrc), "r"(usrc), "r"(vsrc), "g" (chromWidth)
+ : "%"REG_a
+ );
#else
#if defined ARCH_ALPHA && defined HAVE_MVI
-#define pl2yuy2(n) \
- y1 = yc[n]; \
- y2 = yc2[n]; \
- u = uc[n]; \
- v = vc[n]; \
- asm("unpkbw %1, %0" : "=r"(y1) : "r"(y1)); \
- asm("unpkbw %1, %0" : "=r"(y2) : "r"(y2)); \
- asm("unpkbl %1, %0" : "=r"(u) : "r"(u)); \
- asm("unpkbl %1, %0" : "=r"(v) : "r"(v)); \
- yuv1 = (u << 8) + (v << 24); \
- yuv2 = yuv1 + y2; \
- yuv1 += y1; \
- qdst[n] = yuv1; \
- qdst2[n] = yuv2;
-
- int i;
- uint64_t *qdst = (uint64_t *) dst;
- uint64_t *qdst2 = (uint64_t *) (dst + dstStride);
- const uint32_t *yc = (uint32_t *) ysrc;
- const uint32_t *yc2 = (uint32_t *) (ysrc + lumStride);
- const uint16_t *uc = (uint16_t*) usrc, *vc = (uint16_t*) vsrc;
- for(i = 0; i < chromWidth; i += 8){
- uint64_t y1, y2, yuv1, yuv2;
- uint64_t u, v;
- /* Prefetch */
- asm("ldq $31,64(%0)" :: "r"(yc));
- asm("ldq $31,64(%0)" :: "r"(yc2));
- asm("ldq $31,64(%0)" :: "r"(uc));
- asm("ldq $31,64(%0)" :: "r"(vc));
-
- pl2yuy2(0);
- pl2yuy2(1);
- pl2yuy2(2);
- pl2yuy2(3);
-
- yc += 4;
- yc2 += 4;
- uc += 4;
- vc += 4;
- qdst += 4;
- qdst2 += 4;
- }
- y++;
- ysrc += lumStride;
- dst += dstStride;
+#define pl2yuy2(n) \
+ y1 = yc[n]; \
+ y2 = yc2[n]; \
+ u = uc[n]; \
+ v = vc[n]; \
+ asm("unpkbw %1, %0" : "=r"(y1) : "r"(y1)); \
+ asm("unpkbw %1, %0" : "=r"(y2) : "r"(y2)); \
+ asm("unpkbl %1, %0" : "=r"(u) : "r"(u)); \
+ asm("unpkbl %1, %0" : "=r"(v) : "r"(v)); \
+ yuv1 = (u << 8) + (v << 24); \
+ yuv2 = yuv1 + y2; \
+ yuv1 += y1; \
+ qdst[n] = yuv1; \
+ qdst2[n] = yuv2;
+
+ int i;
+ uint64_t *qdst = (uint64_t *) dst;
+ uint64_t *qdst2 = (uint64_t *) (dst + dstStride);
+ const uint32_t *yc = (uint32_t *) ysrc;
+ const uint32_t *yc2 = (uint32_t *) (ysrc + lumStride);
+ const uint16_t *uc = (uint16_t*) usrc, *vc = (uint16_t*) vsrc;
+ for (i = 0; i < chromWidth; i += 8){
+ uint64_t y1, y2, yuv1, yuv2;
+ uint64_t u, v;
+ /* Prefetch */
+ asm("ldq $31,64(%0)" :: "r"(yc));
+ asm("ldq $31,64(%0)" :: "r"(yc2));
+ asm("ldq $31,64(%0)" :: "r"(uc));
+ asm("ldq $31,64(%0)" :: "r"(vc));
+
+ pl2yuy2(0);
+ pl2yuy2(1);
+ pl2yuy2(2);
+ pl2yuy2(3);
+
+ yc += 4;
+ yc2 += 4;
+ uc += 4;
+ vc += 4;
+ qdst += 4;
+ qdst2 += 4;
+ }
+ y++;
+ ysrc += lumStride;
+ dst += dstStride;
#elif __WORDSIZE >= 64
- int i;
- uint64_t *ldst = (uint64_t *) dst;
- const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
- for(i = 0; i < chromWidth; i += 2){
- uint64_t k, l;
- k = yc[0] + (uc[0] << 8) +
- (yc[1] << 16) + (vc[0] << 24);
- l = yc[2] + (uc[1] << 8) +
- (yc[3] << 16) + (vc[1] << 24);
- *ldst++ = k + (l << 32);
- yc += 4;
- uc += 2;
- vc += 2;
- }
+ int i;
+ uint64_t *ldst = (uint64_t *) dst;
+ const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
+ for (i = 0; i < chromWidth; i += 2){
+ uint64_t k, l;
+ k = yc[0] + (uc[0] << 8) +
+ (yc[1] << 16) + (vc[0] << 24);
+ l = yc[2] + (uc[1] << 8) +
+ (yc[3] << 16) + (vc[1] << 24);
+ *ldst++ = k + (l << 32);
+ yc += 4;
+ uc += 2;
+ vc += 2;
+ }
#else
- int i, *idst = (int32_t *) dst;
- const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
- for(i = 0; i < chromWidth; i++){
+ int i, *idst = (int32_t *) dst;
+ const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
+ for (i = 0; i < chromWidth; i++){
#ifdef WORDS_BIGENDIAN
- *idst++ = (yc[0] << 24)+ (uc[0] << 16) +
- (yc[1] << 8) + (vc[0] << 0);
+ *idst++ = (yc[0] << 24)+ (uc[0] << 16) +
+ (yc[1] << 8) + (vc[0] << 0);
#else
- *idst++ = yc[0] + (uc[0] << 8) +
- (yc[1] << 16) + (vc[0] << 24);
-#endif
- yc += 2;
- uc++;
- vc++;
- }
-#endif
-#endif
- if((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
- {
- usrc += chromStride;
- vsrc += chromStride;
- }
- ysrc += lumStride;
- dst += dstStride;
- }
+ *idst++ = yc[0] + (uc[0] << 8) +
+ (yc[1] << 16) + (vc[0] << 24);
+#endif
+ yc += 2;
+ uc++;
+ vc++;
+ }
+#endif
+#endif
+ if ((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
+ {
+ usrc += chromStride;
+ vsrc += chromStride;
+ }
+ ysrc += lumStride;
+ dst += dstStride;
+ }
#ifdef HAVE_MMX
-asm( EMMS" \n\t"
- SFENCE" \n\t"
+asm( EMMS" \n\t"
+ SFENCE" \n\t"
:::"memory");
#endif
}
@@ -1645,103 +1644,103 @@ asm( EMMS" \n\t"
* problem for anyone then tell me, and ill fix it)
*/
static inline void RENAME(yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride)
+ long width, long height,
+ long lumStride, long chromStride, long dstStride)
{
- //FIXME interpolate chroma
- RENAME(yuvPlanartoyuy2)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 2);
+ //FIXME interpolate chroma
+ RENAME(yuvPlanartoyuy2)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 2);
}
static inline void RENAME(yuvPlanartouyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride, long vertLumPerChroma)
+ long width, long height,
+ long lumStride, long chromStride, long dstStride, long vertLumPerChroma)
{
- long y;
- const long chromWidth= width>>1;
- for(y=0; y<height; y++)
- {
+ long y;
+ const long chromWidth= width>>1;
+ for (y=0; y<height; y++)
+ {
#ifdef HAVE_MMX
//FIXME handle 2 lines a once (fewer prefetch, reuse some chrom, but very likely limited by mem anyway)
- asm volatile(
- "xor %%"REG_a", %%"REG_a" \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
- PREFETCH" 32(%2, %%"REG_a") \n\t"
- PREFETCH" 32(%3, %%"REG_a") \n\t"
- "movq (%2, %%"REG_a"), %%mm0 \n\t" // U(0)
- "movq %%mm0, %%mm2 \n\t" // U(0)
- "movq (%3, %%"REG_a"), %%mm1 \n\t" // V(0)
- "punpcklbw %%mm1, %%mm0 \n\t" // UVUV UVUV(0)
- "punpckhbw %%mm1, %%mm2 \n\t" // UVUV UVUV(8)
-
- "movq (%1, %%"REG_a",2), %%mm3 \n\t" // Y(0)
- "movq 8(%1, %%"REG_a",2), %%mm5 \n\t" // Y(8)
- "movq %%mm0, %%mm4 \n\t" // Y(0)
- "movq %%mm2, %%mm6 \n\t" // Y(8)
- "punpcklbw %%mm3, %%mm0 \n\t" // YUYV YUYV(0)
- "punpckhbw %%mm3, %%mm4 \n\t" // YUYV YUYV(4)
- "punpcklbw %%mm5, %%mm2 \n\t" // YUYV YUYV(8)
- "punpckhbw %%mm5, %%mm6 \n\t" // YUYV YUYV(12)
-
- MOVNTQ" %%mm0, (%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm4, 8(%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm2, 16(%0, %%"REG_a", 4)\n\t"
- MOVNTQ" %%mm6, 24(%0, %%"REG_a", 4)\n\t"
-
- "add $8, %%"REG_a" \n\t"
- "cmp %4, %%"REG_a" \n\t"
- " jb 1b \n\t"
- ::"r"(dst), "r"(ysrc), "r"(usrc), "r"(vsrc), "g" (chromWidth)
- : "%"REG_a
- );
+ asm volatile(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 32(%1, %%"REG_a", 2) \n\t"
+ PREFETCH" 32(%2, %%"REG_a") \n\t"
+ PREFETCH" 32(%3, %%"REG_a") \n\t"
+ "movq (%2, %%"REG_a"), %%mm0 \n\t" // U(0)
+ "movq %%mm0, %%mm2 \n\t" // U(0)
+ "movq (%3, %%"REG_a"), %%mm1 \n\t" // V(0)
+ "punpcklbw %%mm1, %%mm0 \n\t" // UVUV UVUV(0)
+ "punpckhbw %%mm1, %%mm2 \n\t" // UVUV UVUV(8)
+
+ "movq (%1, %%"REG_a",2), %%mm3 \n\t" // Y(0)
+ "movq 8(%1, %%"REG_a",2), %%mm5 \n\t" // Y(8)
+ "movq %%mm0, %%mm4 \n\t" // Y(0)
+ "movq %%mm2, %%mm6 \n\t" // Y(8)
+ "punpcklbw %%mm3, %%mm0 \n\t" // YUYV YUYV(0)
+ "punpckhbw %%mm3, %%mm4 \n\t" // YUYV YUYV(4)
+ "punpcklbw %%mm5, %%mm2 \n\t" // YUYV YUYV(8)
+ "punpckhbw %%mm5, %%mm6 \n\t" // YUYV YUYV(12)
+
+ MOVNTQ" %%mm0, (%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm4, 8(%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm2, 16(%0, %%"REG_a", 4) \n\t"
+ MOVNTQ" %%mm6, 24(%0, %%"REG_a", 4) \n\t"
+
+ "add $8, %%"REG_a" \n\t"
+ "cmp %4, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+ ::"r"(dst), "r"(ysrc), "r"(usrc), "r"(vsrc), "g" (chromWidth)
+ : "%"REG_a
+ );
#else
//FIXME adapt the alpha asm code from yv12->yuy2
#if __WORDSIZE >= 64
- int i;
- uint64_t *ldst = (uint64_t *) dst;
- const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
- for(i = 0; i < chromWidth; i += 2){
- uint64_t k, l;
- k = uc[0] + (yc[0] << 8) +
- (vc[0] << 16) + (yc[1] << 24);
- l = uc[1] + (yc[2] << 8) +
- (vc[1] << 16) + (yc[3] << 24);
- *ldst++ = k + (l << 32);
- yc += 4;
- uc += 2;
- vc += 2;
- }
+ int i;
+ uint64_t *ldst = (uint64_t *) dst;
+ const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
+ for (i = 0; i < chromWidth; i += 2){
+ uint64_t k, l;
+ k = uc[0] + (yc[0] << 8) +
+ (vc[0] << 16) + (yc[1] << 24);
+ l = uc[1] + (yc[2] << 8) +
+ (vc[1] << 16) + (yc[3] << 24);
+ *ldst++ = k + (l << 32);
+ yc += 4;
+ uc += 2;
+ vc += 2;
+ }
#else
- int i, *idst = (int32_t *) dst;
- const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
- for(i = 0; i < chromWidth; i++){
+ int i, *idst = (int32_t *) dst;
+ const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
+ for (i = 0; i < chromWidth; i++){
#ifdef WORDS_BIGENDIAN
- *idst++ = (uc[0] << 24)+ (yc[0] << 16) +
- (vc[0] << 8) + (yc[1] << 0);
+ *idst++ = (uc[0] << 24)+ (yc[0] << 16) +
+ (vc[0] << 8) + (yc[1] << 0);
#else
- *idst++ = uc[0] + (yc[0] << 8) +
- (vc[0] << 16) + (yc[1] << 24);
-#endif
- yc += 2;
- uc++;
- vc++;
- }
-#endif
-#endif
- if((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
- {
- usrc += chromStride;
- vsrc += chromStride;
- }
- ysrc += lumStride;
- dst += dstStride;
- }
+ *idst++ = uc[0] + (yc[0] << 8) +
+ (vc[0] << 16) + (yc[1] << 24);
+#endif
+ yc += 2;
+ uc++;
+ vc++;
+ }
+#endif
+#endif
+ if ((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
+ {
+ usrc += chromStride;
+ vsrc += chromStride;
+ }
+ ysrc += lumStride;
+ dst += dstStride;
+ }
#ifdef HAVE_MMX
-asm( EMMS" \n\t"
- SFENCE" \n\t"
+asm( EMMS" \n\t"
+ SFENCE" \n\t"
:::"memory");
#endif
}
@@ -1752,11 +1751,11 @@ asm( EMMS" \n\t"
* problem for anyone then tell me, and ill fix it)
*/
static inline void RENAME(yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride)
+ long width, long height,
+ long lumStride, long chromStride, long dstStride)
{
- //FIXME interpolate chroma
- RENAME(yuvPlanartouyvy)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 2);
+ //FIXME interpolate chroma
+ RENAME(yuvPlanartouyvy)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 2);
}
/**
@@ -1764,10 +1763,10 @@ static inline void RENAME(yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc,
* width should be a multiple of 16
*/
static inline void RENAME(yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
- long width, long height,
- long lumStride, long chromStride, long dstStride)
+ long width, long height,
+ long lumStride, long chromStride, long dstStride)
{
- RENAME(yuvPlanartoyuy2)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 1);
+ RENAME(yuvPlanartoyuy2)(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride, 1);
}
/**
@@ -1776,234 +1775,234 @@ static inline void RENAME(yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usr
* problem for anyone then tell me, and ill fix it)
*/
static inline void RENAME(yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height,
- long lumStride, long chromStride, long srcStride)
+ long width, long height,
+ long lumStride, long chromStride, long srcStride)
{
- long y;
- const long chromWidth= width>>1;
- for(y=0; y<height; y+=2)
- {
+ long y;
+ const long chromWidth= width>>1;
+ for (y=0; y<height; y+=2)
+ {
#ifdef HAVE_MMX
- asm volatile(
- "xor %%"REG_a", %%"REG_a" \n\t"
- "pcmpeqw %%mm7, %%mm7 \n\t"
- "psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
- "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
- "movq 8(%0, %%"REG_a", 4), %%mm1\n\t" // YUYV YUYV(4)
- "movq %%mm0, %%mm2 \n\t" // YUYV YUYV(0)
- "movq %%mm1, %%mm3 \n\t" // YUYV YUYV(4)
- "psrlw $8, %%mm0 \n\t" // U0V0 U0V0(0)
- "psrlw $8, %%mm1 \n\t" // U0V0 U0V0(4)
- "pand %%mm7, %%mm2 \n\t" // Y0Y0 Y0Y0(0)
- "pand %%mm7, %%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, %%"REG_a", 2)\n\t"
-
- "movq 16(%0, %%"REG_a", 4), %%mm1\n\t" // YUYV YUYV(8)
- "movq 24(%0, %%"REG_a", 4), %%mm2\n\t" // YUYV YUYV(12)
- "movq %%mm1, %%mm3 \n\t" // YUYV YUYV(8)
- "movq %%mm2, %%mm4 \n\t" // YUYV YUYV(12)
- "psrlw $8, %%mm1 \n\t" // U0V0 U0V0(8)
- "psrlw $8, %%mm2 \n\t" // U0V0 U0V0(12)
- "pand %%mm7, %%mm3 \n\t" // Y0Y0 Y0Y0(8)
- "pand %%mm7, %%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, %%"REG_a", 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, %%"REG_a") \n\t"
- MOVNTQ" %%mm2, (%2, %%"REG_a") \n\t"
-
- "add $8, %%"REG_a" \n\t"
- "cmp %4, %%"REG_a" \n\t"
- " jb 1b \n\t"
- ::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
- : "memory", "%"REG_a
- );
-
- ydst += lumStride;
- src += srcStride;
-
- asm volatile(
- "xor %%"REG_a", %%"REG_a" \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
- "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
- "movq 8(%0, %%"REG_a", 4), %%mm1\n\t" // YUYV YUYV(4)
- "movq 16(%0, %%"REG_a", 4), %%mm2\n\t" // YUYV YUYV(8)
- "movq 24(%0, %%"REG_a", 4), %%mm3\n\t" // YUYV YUYV(12)
- "pand %%mm7, %%mm0 \n\t" // Y0Y0 Y0Y0(0)
- "pand %%mm7, %%mm1 \n\t" // Y0Y0 Y0Y0(4)
- "pand %%mm7, %%mm2 \n\t" // Y0Y0 Y0Y0(8)
- "pand %%mm7, %%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, %%"REG_a", 2)\n\t"
- MOVNTQ" %%mm2, 8(%1, %%"REG_a", 2)\n\t"
-
- "add $8, %%"REG_a" \n\t"
- "cmp %4, %%"REG_a" \n\t"
- " jb 1b \n\t"
-
- ::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
- : "memory", "%"REG_a
- );
+ asm volatile(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ "pcmpeqw %%mm7, %%mm7 \n\t"
+ "psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
+ "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
+ "movq 8(%0, %%"REG_a", 4), %%mm1 \n\t" // YUYV YUYV(4)
+ "movq %%mm0, %%mm2 \n\t" // YUYV YUYV(0)
+ "movq %%mm1, %%mm3 \n\t" // YUYV YUYV(4)
+ "psrlw $8, %%mm0 \n\t" // U0V0 U0V0(0)
+ "psrlw $8, %%mm1 \n\t" // U0V0 U0V0(4)
+ "pand %%mm7, %%mm2 \n\t" // Y0Y0 Y0Y0(0)
+ "pand %%mm7, %%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, %%"REG_a", 2) \n\t"
+
+ "movq 16(%0, %%"REG_a", 4), %%mm1 \n\t" // YUYV YUYV(8)
+ "movq 24(%0, %%"REG_a", 4), %%mm2 \n\t" // YUYV YUYV(12)
+ "movq %%mm1, %%mm3 \n\t" // YUYV YUYV(8)
+ "movq %%mm2, %%mm4 \n\t" // YUYV YUYV(12)
+ "psrlw $8, %%mm1 \n\t" // U0V0 U0V0(8)
+ "psrlw $8, %%mm2 \n\t" // U0V0 U0V0(12)
+ "pand %%mm7, %%mm3 \n\t" // Y0Y0 Y0Y0(8)
+ "pand %%mm7, %%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, %%"REG_a", 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, %%"REG_a") \n\t"
+ MOVNTQ" %%mm2, (%2, %%"REG_a") \n\t"
+
+ "add $8, %%"REG_a" \n\t"
+ "cmp %4, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+ ::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
+ : "memory", "%"REG_a
+ );
+
+ ydst += lumStride;
+ src += srcStride;
+
+ asm volatile(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 64(%0, %%"REG_a", 4) \n\t"
+ "movq (%0, %%"REG_a", 4), %%mm0 \n\t" // YUYV YUYV(0)
+ "movq 8(%0, %%"REG_a", 4), %%mm1 \n\t" // YUYV YUYV(4)
+ "movq 16(%0, %%"REG_a", 4), %%mm2 \n\t" // YUYV YUYV(8)
+ "movq 24(%0, %%"REG_a", 4), %%mm3 \n\t" // YUYV YUYV(12)
+ "pand %%mm7, %%mm0 \n\t" // Y0Y0 Y0Y0(0)
+ "pand %%mm7, %%mm1 \n\t" // Y0Y0 Y0Y0(4)
+ "pand %%mm7, %%mm2 \n\t" // Y0Y0 Y0Y0(8)
+ "pand %%mm7, %%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, %%"REG_a", 2) \n\t"
+ MOVNTQ" %%mm2, 8(%1, %%"REG_a", 2) \n\t"
+
+ "add $8, %%"REG_a" \n\t"
+ "cmp %4, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+
+ ::"r"(src), "r"(ydst), "r"(udst), "r"(vdst), "g" (chromWidth)
+ : "memory", "%"REG_a
+ );
#else
- long i;
- for(i=0; i<chromWidth; i++)
- {
- ydst[2*i+0] = src[4*i+0];
- udst[i] = src[4*i+1];
- ydst[2*i+1] = src[4*i+2];
- vdst[i] = src[4*i+3];
- }
- ydst += lumStride;
- src += srcStride;
-
- for(i=0; i<chromWidth; i++)
- {
- ydst[2*i+0] = src[4*i+0];
- ydst[2*i+1] = src[4*i+2];
- }
-#endif
- udst += chromStride;
- vdst += chromStride;
- ydst += lumStride;
- src += srcStride;
- }
+ long i;
+ for (i=0; i<chromWidth; i++)
+ {
+ ydst[2*i+0] = src[4*i+0];
+ udst[i] = src[4*i+1];
+ ydst[2*i+1] = src[4*i+2];
+ vdst[i] = src[4*i+3];
+ }
+ ydst += lumStride;
+ src += srcStride;
+
+ for (i=0; i<chromWidth; i++)
+ {
+ ydst[2*i+0] = src[4*i+0];
+ ydst[2*i+1] = src[4*i+2];
+ }
+#endif
+ udst += chromStride;
+ vdst += chromStride;
+ ydst += lumStride;
+ src += srcStride;
+ }
#ifdef HAVE_MMX
-asm volatile( EMMS" \n\t"
- SFENCE" \n\t"
- :::"memory");
+asm volatile( EMMS" \n\t"
+ SFENCE" \n\t"
+ :::"memory");
#endif
}
static inline void RENAME(yvu9toyv12)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc,
- uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
- long width, long height, long lumStride, long chromStride)
+ uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
+ long width, long height, long lumStride, long chromStride)
{
- /* Y Plane */
- memcpy(ydst, ysrc, width*height);
+ /* Y Plane */
+ memcpy(ydst, ysrc, width*height);
- /* XXX: implement upscaling for U,V */
+ /* XXX: implement upscaling for U,V */
}
static inline void RENAME(planar2x)(const uint8_t *src, uint8_t *dst, long srcWidth, long srcHeight, long srcStride, long dstStride)
{
- long x,y;
+ long x,y;
- dst[0]= src[0];
+ dst[0]= src[0];
- // first line
- 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];
+ // first line
+ 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];
dst+= dstStride;
- for(y=1; y<srcHeight; y++){
+ for (y=1; y<srcHeight; y++){
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
- const long mmxSize= srcWidth&~15;
- asm volatile(
- "mov %4, %%"REG_a" \n\t"
- "1: \n\t"
- "movq (%0, %%"REG_a"), %%mm0 \n\t"
- "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"
+ const long mmxSize= srcWidth&~15;
+ asm volatile(
+ "mov %4, %%"REG_a" \n\t"
+ "1: \n\t"
+ "movq (%0, %%"REG_a"), %%mm0 \n\t"
+ "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"
+ 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
-
- );
+ "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;
- }
+ 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
+ // last line
#if 1
- dst[0]= src[0];
+ 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];
+ 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];
- }
+ 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");
+asm volatile( EMMS" \n\t"
+ SFENCE" \n\t"
+ :::"memory");
#endif
}
@@ -2014,122 +2013,122 @@ asm volatile( EMMS" \n\t"
* 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,
- long lumStride, long chromStride, long srcStride)
+ long width, long height,
+ long lumStride, long chromStride, long srcStride)
{
- long y;
- const long chromWidth= width>>1;
- for(y=0; y<height; y+=2)
- {
+ 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...
- ASMALIGN(4)
- "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"
- ASMALIGN(4)
- "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"
- );
+ asm volatile(
+ "xorl %%eax, %%eax \n\t"
+ "pcmpeqw %%mm7, %%mm7 \n\t"
+ "psrlw $8, %%mm7 \n\t" // FF,00,FF,00...
+ ASMALIGN(4)
+ "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"
+ ASMALIGN(4)
+ "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;
- }
+ 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");
+asm volatile( EMMS" \n\t"
+ SFENCE" \n\t"
+ :::"memory");
#endif
}
@@ -2140,604 +2139,604 @@ asm volatile( EMMS" \n\t"
* 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,
- long lumStride, long chromStride, long srcStride)
+ long width, long height,
+ long lumStride, long chromStride, long srcStride)
{
- long y;
- const long chromWidth= width>>1;
+ 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_d"\n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 64(%0, %%"REG_d") \n\t"
- "movd (%0, %%"REG_d"), %%mm0 \n\t"
- "movd 3(%0, %%"REG_d"), %%mm1 \n\t"
- "punpcklbw %%mm7, %%mm0 \n\t"
- "punpcklbw %%mm7, %%mm1 \n\t"
- "movd 6(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 9(%0, %%"REG_d"), %%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"
+ 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_d" \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 64(%0, %%"REG_d") \n\t"
+ "movd (%0, %%"REG_d"), %%mm0 \n\t"
+ "movd 3(%0, %%"REG_d"), %%mm1 \n\t"
+ "punpcklbw %%mm7, %%mm0 \n\t"
+ "punpcklbw %%mm7, %%mm1 \n\t"
+ "movd 6(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 9(%0, %%"REG_d"), %%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_d"), %%mm4 \n\t"
- "movd 15(%0, %%"REG_d"), %%mm1 \n\t"
- "punpcklbw %%mm7, %%mm4 \n\t"
- "punpcklbw %%mm7, %%mm1 \n\t"
- "movd 18(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 21(%0, %%"REG_d"), %%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"
+ "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_d"), %%mm4 \n\t"
+ "movd 15(%0, %%"REG_d"), %%mm1 \n\t"
+ "punpcklbw %%mm7, %%mm4 \n\t"
+ "punpcklbw %%mm7, %%mm1 \n\t"
+ "movd 18(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 21(%0, %%"REG_d"), %%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_d" \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_d
- );
- 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_d"\n\t"
- "add %%"REG_d", %%"REG_d" \n\t"
- ASMALIGN(4)
- "1: \n\t"
- PREFETCH" 64(%0, %%"REG_d") \n\t"
- PREFETCH" 64(%1, %%"REG_d") \n\t"
+ "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_d" \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_d
+ );
+ 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_d" \n\t"
+ "add %%"REG_d", %%"REG_d" \n\t"
+ ASMALIGN(4)
+ "1: \n\t"
+ PREFETCH" 64(%0, %%"REG_d") \n\t"
+ PREFETCH" 64(%1, %%"REG_d") \n\t"
#if defined (HAVE_MMX2) || defined (HAVE_3DNOW)
- "movq (%0, %%"REG_d"), %%mm0 \n\t"
- "movq (%1, %%"REG_d"), %%mm1 \n\t"
- "movq 6(%0, %%"REG_d"), %%mm2 \n\t"
- "movq 6(%1, %%"REG_d"), %%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"
+ "movq (%0, %%"REG_d"), %%mm0 \n\t"
+ "movq (%1, %%"REG_d"), %%mm1 \n\t"
+ "movq 6(%0, %%"REG_d"), %%mm2 \n\t"
+ "movq 6(%1, %%"REG_d"), %%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_d"), %%mm0 \n\t"
- "movd (%1, %%"REG_d"), %%mm1 \n\t"
- "movd 3(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 3(%1, %%"REG_d"), %%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_d"), %%mm4 \n\t"
- "movd 6(%1, %%"REG_d"), %%mm1 \n\t"
- "movd 9(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 9(%1, %%"REG_d"), %%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"
+ "movd (%0, %%"REG_d"), %%mm0 \n\t"
+ "movd (%1, %%"REG_d"), %%mm1 \n\t"
+ "movd 3(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 3(%1, %%"REG_d"), %%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_d"), %%mm4 \n\t"
+ "movd 6(%1, %%"REG_d"), %%mm1 \n\t"
+ "movd 9(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 9(%1, %%"REG_d"), %%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"
+ "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_d"), %%mm4 \n\t"
- "movq 12(%1, %%"REG_d"), %%mm1 \n\t"
- "movq 18(%0, %%"REG_d"), %%mm2 \n\t"
- "movq 18(%1, %%"REG_d"), %%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"
+ "movq 12(%0, %%"REG_d"), %%mm4 \n\t"
+ "movq 12(%1, %%"REG_d"), %%mm1 \n\t"
+ "movq 18(%0, %%"REG_d"), %%mm2 \n\t"
+ "movq 18(%1, %%"REG_d"), %%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_d"), %%mm4 \n\t"
- "movd 12(%1, %%"REG_d"), %%mm1 \n\t"
- "movd 15(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 15(%1, %%"REG_d"), %%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_d"), %%mm5 \n\t"
- "movd 18(%1, %%"REG_d"), %%mm1 \n\t"
- "movd 21(%0, %%"REG_d"), %%mm2 \n\t"
- "movd 21(%1, %%"REG_d"), %%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"
+ "movd 12(%0, %%"REG_d"), %%mm4 \n\t"
+ "movd 12(%1, %%"REG_d"), %%mm1 \n\t"
+ "movd 15(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 15(%1, %%"REG_d"), %%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_d"), %%mm5 \n\t"
+ "movd 18(%1, %%"REG_d"), %%mm1 \n\t"
+ "movd 21(%0, %%"REG_d"), %%mm2 \n\t"
+ "movd 21(%1, %%"REG_d"), %%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_d" \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_d
- );
-
- udst += chromStride;
- vdst += chromStride;
- src += srcStride*2;
- }
-
- asm volatile( EMMS" \n\t"
- SFENCE" \n\t"
- :::"memory");
+ "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_d" \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_d
+ );
+
+ 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 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;
-
- 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;
- }
- udst += chromStride;
- vdst += chromStride;
- ydst += lumStride;
- src += srcStride;
- }
+ 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 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;
+
+ 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;
+ }
+ udst += chromStride;
+ vdst += chromStride;
+ ydst += lumStride;
+ src += srcStride;
+ }
}
void RENAME(interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dest,
- long width, long height, long src1Stride,
- long src2Stride, long dstStride){
- long h;
+ long width, long height, long src1Stride,
+ long src2Stride, long dstStride){
+ long h;
- for(h=0; h < height; h++)
- {
- long w;
+ for (h=0; h < height; h++)
+ {
+ long w;
#ifdef HAVE_MMX
#ifdef HAVE_SSE2
- asm(
- "xor %%"REG_a", %%"REG_a" \n\t"
- "1: \n\t"
- PREFETCH" 64(%1, %%"REG_a") \n\t"
- PREFETCH" 64(%2, %%"REG_a") \n\t"
- "movdqa (%1, %%"REG_a"), %%xmm0 \n\t"
- "movdqa (%1, %%"REG_a"), %%xmm1 \n\t"
- "movdqa (%2, %%"REG_a"), %%xmm2 \n\t"
- "punpcklbw %%xmm2, %%xmm0 \n\t"
- "punpckhbw %%xmm2, %%xmm1 \n\t"
- "movntdq %%xmm0, (%0, %%"REG_a", 2)\n\t"
- "movntdq %%xmm1, 16(%0, %%"REG_a", 2)\n\t"
- "add $16, %%"REG_a" \n\t"
- "cmp %3, %%"REG_a" \n\t"
- " jb 1b \n\t"
- ::"r"(dest), "r"(src1), "r"(src2), "r" (width-15)
- : "memory", "%"REG_a""
- );
+ asm(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ "1: \n\t"
+ PREFETCH" 64(%1, %%"REG_a") \n\t"
+ PREFETCH" 64(%2, %%"REG_a") \n\t"
+ "movdqa (%1, %%"REG_a"), %%xmm0 \n\t"
+ "movdqa (%1, %%"REG_a"), %%xmm1 \n\t"
+ "movdqa (%2, %%"REG_a"), %%xmm2 \n\t"
+ "punpcklbw %%xmm2, %%xmm0 \n\t"
+ "punpckhbw %%xmm2, %%xmm1 \n\t"
+ "movntdq %%xmm0, (%0, %%"REG_a", 2) \n\t"
+ "movntdq %%xmm1, 16(%0, %%"REG_a", 2) \n\t"
+ "add $16, %%"REG_a" \n\t"
+ "cmp %3, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+ ::"r"(dest), "r"(src1), "r"(src2), "r" (width-15)
+ : "memory", "%"REG_a""
+ );
#else
- asm(
- "xor %%"REG_a", %%"REG_a" \n\t"
- "1: \n\t"
- PREFETCH" 64(%1, %%"REG_a") \n\t"
- PREFETCH" 64(%2, %%"REG_a") \n\t"
- "movq (%1, %%"REG_a"), %%mm0 \n\t"
- "movq 8(%1, %%"REG_a"), %%mm2 \n\t"
- "movq %%mm0, %%mm1 \n\t"
- "movq %%mm2, %%mm3 \n\t"
- "movq (%2, %%"REG_a"), %%mm4 \n\t"
- "movq 8(%2, %%"REG_a"), %%mm5 \n\t"
- "punpcklbw %%mm4, %%mm0 \n\t"
- "punpckhbw %%mm4, %%mm1 \n\t"
- "punpcklbw %%mm5, %%mm2 \n\t"
- "punpckhbw %%mm5, %%mm3 \n\t"
- MOVNTQ" %%mm0, (%0, %%"REG_a", 2)\n\t"
- MOVNTQ" %%mm1, 8(%0, %%"REG_a", 2)\n\t"
- MOVNTQ" %%mm2, 16(%0, %%"REG_a", 2)\n\t"
- MOVNTQ" %%mm3, 24(%0, %%"REG_a", 2)\n\t"
- "add $16, %%"REG_a" \n\t"
- "cmp %3, %%"REG_a" \n\t"
- " jb 1b \n\t"
- ::"r"(dest), "r"(src1), "r"(src2), "r" (width-15)
- : "memory", "%"REG_a
- );
-#endif
- for(w= (width&(~15)); w < width; w++)
- {
- dest[2*w+0] = src1[w];
- dest[2*w+1] = src2[w];
- }
+ asm(
+ "xor %%"REG_a", %%"REG_a" \n\t"
+ "1: \n\t"
+ PREFETCH" 64(%1, %%"REG_a") \n\t"
+ PREFETCH" 64(%2, %%"REG_a") \n\t"
+ "movq (%1, %%"REG_a"), %%mm0 \n\t"
+ "movq 8(%1, %%"REG_a"), %%mm2 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "movq (%2, %%"REG_a"), %%mm4 \n\t"
+ "movq 8(%2, %%"REG_a"), %%mm5 \n\t"
+ "punpcklbw %%mm4, %%mm0 \n\t"
+ "punpckhbw %%mm4, %%mm1 \n\t"
+ "punpcklbw %%mm5, %%mm2 \n\t"
+ "punpckhbw %%mm5, %%mm3 \n\t"
+ MOVNTQ" %%mm0, (%0, %%"REG_a", 2) \n\t"
+ MOVNTQ" %%mm1, 8(%0, %%"REG_a", 2) \n\t"
+ MOVNTQ" %%mm2, 16(%0, %%"REG_a", 2) \n\t"
+ MOVNTQ" %%mm3, 24(%0, %%"REG_a", 2) \n\t"
+ "add $16, %%"REG_a" \n\t"
+ "cmp %3, %%"REG_a" \n\t"
+ " jb 1b \n\t"
+ ::"r"(dest), "r"(src1), "r"(src2), "r" (width-15)
+ : "memory", "%"REG_a
+ );
+#endif
+ for (w= (width&(~15)); w < width; w++)
+ {
+ dest[2*w+0] = src1[w];
+ dest[2*w+1] = src2[w];
+ }
#else
- for(w=0; w < width; w++)
- {
- dest[2*w+0] = src1[w];
- dest[2*w+1] = src2[w];
- }
+ for (w=0; w < width; w++)
+ {
+ dest[2*w+0] = src1[w];
+ dest[2*w+1] = src2[w];
+ }
#endif
- dest += dstStride;
+ dest += dstStride;
src1 += src1Stride;
src2 += src2Stride;
- }
+ }
#ifdef HAVE_MMX
- asm(
- EMMS" \n\t"
- SFENCE" \n\t"
- ::: "memory"
- );
+ asm(
+ EMMS" \n\t"
+ SFENCE" \n\t"
+ ::: "memory"
+ );
#endif
}
static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
- uint8_t *dst1, uint8_t *dst2,
- long width, long height,
- long srcStride1, long srcStride2,
- long dstStride1, long dstStride2)
+ uint8_t *dst1, uint8_t *dst2,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long dstStride1, long dstStride2)
{
long y,x,w,h;
w=width/2; h=height/2;
#ifdef HAVE_MMX
asm volatile(
- PREFETCH" %0\n\t"
- PREFETCH" %1\n\t"
- ::"m"(*(src1+srcStride1)),"m"(*(src2+srcStride2)):"memory");
-#endif
- for(y=0;y<h;y++){
- const uint8_t* s1=src1+srcStride1*(y>>1);
- uint8_t* d=dst1+dstStride1*y;
- x=0;
+ PREFETCH" %0 \n\t"
+ PREFETCH" %1 \n\t"
+ ::"m"(*(src1+srcStride1)),"m"(*(src2+srcStride2)):"memory");
+#endif
+ for (y=0;y<h;y++){
+ const uint8_t* s1=src1+srcStride1*(y>>1);
+ uint8_t* d=dst1+dstStride1*y;
+ x=0;
#ifdef HAVE_MMX
- for(;x<w-31;x+=32)
- {
- asm volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq 8%1, %%mm2\n\t"
- "movq 16%1, %%mm4\n\t"
- "movq 24%1, %%mm6\n\t"
- "movq %%mm0, %%mm1\n\t"
- "movq %%mm2, %%mm3\n\t"
- "movq %%mm4, %%mm5\n\t"
- "movq %%mm6, %%mm7\n\t"
- "punpcklbw %%mm0, %%mm0\n\t"
- "punpckhbw %%mm1, %%mm1\n\t"
- "punpcklbw %%mm2, %%mm2\n\t"
- "punpckhbw %%mm3, %%mm3\n\t"
- "punpcklbw %%mm4, %%mm4\n\t"
- "punpckhbw %%mm5, %%mm5\n\t"
- "punpcklbw %%mm6, %%mm6\n\t"
- "punpckhbw %%mm7, %%mm7\n\t"
- MOVNTQ" %%mm0, %0\n\t"
- MOVNTQ" %%mm1, 8%0\n\t"
- MOVNTQ" %%mm2, 16%0\n\t"
- MOVNTQ" %%mm3, 24%0\n\t"
- MOVNTQ" %%mm4, 32%0\n\t"
- MOVNTQ" %%mm5, 40%0\n\t"
- MOVNTQ" %%mm6, 48%0\n\t"
- MOVNTQ" %%mm7, 56%0"
- :"=m"(d[2*x])
- :"m"(s1[x])
- :"memory");
- }
-#endif
- for(;x<w;x++) d[2*x]=d[2*x+1]=s1[x];
+ for (;x<w-31;x+=32)
+ {
+ asm volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "movq 16%1, %%mm4 \n\t"
+ "movq 24%1, %%mm6 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "movq %%mm4, %%mm5 \n\t"
+ "movq %%mm6, %%mm7 \n\t"
+ "punpcklbw %%mm0, %%mm0 \n\t"
+ "punpckhbw %%mm1, %%mm1 \n\t"
+ "punpcklbw %%mm2, %%mm2 \n\t"
+ "punpckhbw %%mm3, %%mm3 \n\t"
+ "punpcklbw %%mm4, %%mm4 \n\t"
+ "punpckhbw %%mm5, %%mm5 \n\t"
+ "punpcklbw %%mm6, %%mm6 \n\t"
+ "punpckhbw %%mm7, %%mm7 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm1, 8%0 \n\t"
+ MOVNTQ" %%mm2, 16%0 \n\t"
+ MOVNTQ" %%mm3, 24%0 \n\t"
+ MOVNTQ" %%mm4, 32%0 \n\t"
+ MOVNTQ" %%mm5, 40%0 \n\t"
+ MOVNTQ" %%mm6, 48%0 \n\t"
+ MOVNTQ" %%mm7, 56%0"
+ :"=m"(d[2*x])
+ :"m"(s1[x])
+ :"memory");
+ }
+#endif
+ for (;x<w;x++) d[2*x]=d[2*x+1]=s1[x];
}
- for(y=0;y<h;y++){
- const uint8_t* s2=src2+srcStride2*(y>>1);
- uint8_t* d=dst2+dstStride2*y;
- x=0;
+ for (y=0;y<h;y++){
+ const uint8_t* s2=src2+srcStride2*(y>>1);
+ uint8_t* d=dst2+dstStride2*y;
+ x=0;
#ifdef HAVE_MMX
- for(;x<w-31;x+=32)
- {
- asm volatile(
- PREFETCH" 32%1\n\t"
- "movq %1, %%mm0\n\t"
- "movq 8%1, %%mm2\n\t"
- "movq 16%1, %%mm4\n\t"
- "movq 24%1, %%mm6\n\t"
- "movq %%mm0, %%mm1\n\t"
- "movq %%mm2, %%mm3\n\t"
- "movq %%mm4, %%mm5\n\t"
- "movq %%mm6, %%mm7\n\t"
- "punpcklbw %%mm0, %%mm0\n\t"
- "punpckhbw %%mm1, %%mm1\n\t"
- "punpcklbw %%mm2, %%mm2\n\t"
- "punpckhbw %%mm3, %%mm3\n\t"
- "punpcklbw %%mm4, %%mm4\n\t"
- "punpckhbw %%mm5, %%mm5\n\t"
- "punpcklbw %%mm6, %%mm6\n\t"
- "punpckhbw %%mm7, %%mm7\n\t"
- MOVNTQ" %%mm0, %0\n\t"
- MOVNTQ" %%mm1, 8%0\n\t"
- MOVNTQ" %%mm2, 16%0\n\t"
- MOVNTQ" %%mm3, 24%0\n\t"
- MOVNTQ" %%mm4, 32%0\n\t"
- MOVNTQ" %%mm5, 40%0\n\t"
- MOVNTQ" %%mm6, 48%0\n\t"
- MOVNTQ" %%mm7, 56%0"
- :"=m"(d[2*x])
- :"m"(s2[x])
- :"memory");
- }
-#endif
- for(;x<w;x++) d[2*x]=d[2*x+1]=s2[x];
+ for (;x<w-31;x+=32)
+ {
+ asm volatile(
+ PREFETCH" 32%1 \n\t"
+ "movq %1, %%mm0 \n\t"
+ "movq 8%1, %%mm2 \n\t"
+ "movq 16%1, %%mm4 \n\t"
+ "movq 24%1, %%mm6 \n\t"
+ "movq %%mm0, %%mm1 \n\t"
+ "movq %%mm2, %%mm3 \n\t"
+ "movq %%mm4, %%mm5 \n\t"
+ "movq %%mm6, %%mm7 \n\t"
+ "punpcklbw %%mm0, %%mm0 \n\t"
+ "punpckhbw %%mm1, %%mm1 \n\t"
+ "punpcklbw %%mm2, %%mm2 \n\t"
+ "punpckhbw %%mm3, %%mm3 \n\t"
+ "punpcklbw %%mm4, %%mm4 \n\t"
+ "punpckhbw %%mm5, %%mm5 \n\t"
+ "punpcklbw %%mm6, %%mm6 \n\t"
+ "punpckhbw %%mm7, %%mm7 \n\t"
+ MOVNTQ" %%mm0, %0 \n\t"
+ MOVNTQ" %%mm1, 8%0 \n\t"
+ MOVNTQ" %%mm2, 16%0 \n\t"
+ MOVNTQ" %%mm3, 24%0 \n\t"
+ MOVNTQ" %%mm4, 32%0 \n\t"
+ MOVNTQ" %%mm5, 40%0 \n\t"
+ MOVNTQ" %%mm6, 48%0 \n\t"
+ MOVNTQ" %%mm7, 56%0"
+ :"=m"(d[2*x])
+ :"m"(s2[x])
+ :"memory");
+ }
+#endif
+ for (;x<w;x++) d[2*x]=d[2*x+1]=s2[x];
}
#ifdef HAVE_MMX
- asm(
- EMMS" \n\t"
- SFENCE" \n\t"
- ::: "memory"
- );
+ asm(
+ EMMS" \n\t"
+ SFENCE" \n\t"
+ ::: "memory"
+ );
#endif
}
static inline void RENAME(yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
- uint8_t *dst,
- long width, long height,
- long srcStride1, long srcStride2,
- long srcStride3, long dstStride)
+ uint8_t *dst,
+ long width, long height,
+ long srcStride1, long srcStride2,
+ long srcStride3, long dstStride)
{
long y,x,w,h;
w=width/2; h=height;
- for(y=0;y<h;y++){
- const uint8_t* yp=src1+srcStride1*y;
- const uint8_t* up=src2+srcStride2*(y>>2);
- const uint8_t* vp=src3+srcStride3*(y>>2);
- uint8_t* d=dst+dstStride*y;
- x=0;
+ for (y=0;y<h;y++){
+ const uint8_t* yp=src1+srcStride1*y;
+ const uint8_t* up=src2+srcStride2*(y>>2);
+ const uint8_t* vp=src3+srcStride3*(y>>2);
+ uint8_t* d=dst+dstStride*y;
+ x=0;
#ifdef HAVE_MMX
- for(;x<w-7;x+=8)
- {
- asm volatile(
- PREFETCH" 32(%1, %0)\n\t"
- PREFETCH" 32(%2, %0)\n\t"
- PREFETCH" 32(%3, %0)\n\t"
- "movq (%1, %0, 4), %%mm0\n\t" /* Y0Y1Y2Y3Y4Y5Y6Y7 */
- "movq (%2, %0), %%mm1\n\t" /* U0U1U2U3U4U5U6U7 */
- "movq (%3, %0), %%mm2\n\t" /* V0V1V2V3V4V5V6V7 */
- "movq %%mm0, %%mm3\n\t" /* Y0Y1Y2Y3Y4Y5Y6Y7 */
- "movq %%mm1, %%mm4\n\t" /* U0U1U2U3U4U5U6U7 */
- "movq %%mm2, %%mm5\n\t" /* V0V1V2V3V4V5V6V7 */
- "punpcklbw %%mm1, %%mm1\n\t" /* U0U0 U1U1 U2U2 U3U3 */
- "punpcklbw %%mm2, %%mm2\n\t" /* V0V0 V1V1 V2V2 V3V3 */
- "punpckhbw %%mm4, %%mm4\n\t" /* U4U4 U5U5 U6U6 U7U7 */
- "punpckhbw %%mm5, %%mm5\n\t" /* V4V4 V5V5 V6V6 V7V7 */
-
- "movq %%mm1, %%mm6\n\t"
- "punpcklbw %%mm2, %%mm1\n\t" /* U0V0 U0V0 U1V1 U1V1*/
- "punpcklbw %%mm1, %%mm0\n\t" /* Y0U0 Y1V0 Y2U0 Y3V0*/
- "punpckhbw %%mm1, %%mm3\n\t" /* Y4U1 Y5V1 Y6U1 Y7V1*/
- MOVNTQ" %%mm0, (%4, %0, 8)\n\t"
- MOVNTQ" %%mm3, 8(%4, %0, 8)\n\t"
-
- "punpckhbw %%mm2, %%mm6\n\t" /* U2V2 U2V2 U3V3 U3V3*/
- "movq 8(%1, %0, 4), %%mm0\n\t"
- "movq %%mm0, %%mm3\n\t"
- "punpcklbw %%mm6, %%mm0\n\t" /* Y U2 Y V2 Y U2 Y V2*/
- "punpckhbw %%mm6, %%mm3\n\t" /* Y U3 Y V3 Y U3 Y V3*/
- MOVNTQ" %%mm0, 16(%4, %0, 8)\n\t"
- MOVNTQ" %%mm3, 24(%4, %0, 8)\n\t"
-
- "movq %%mm4, %%mm6\n\t"
- "movq 16(%1, %0, 4), %%mm0\n\t"
- "movq %%mm0, %%mm3\n\t"
- "punpcklbw %%mm5, %%mm4\n\t"
- "punpcklbw %%mm4, %%mm0\n\t" /* Y U4 Y V4 Y U4 Y V4*/
- "punpckhbw %%mm4, %%mm3\n\t" /* Y U5 Y V5 Y U5 Y V5*/
- MOVNTQ" %%mm0, 32(%4, %0, 8)\n\t"
- MOVNTQ" %%mm3, 40(%4, %0, 8)\n\t"
-
- "punpckhbw %%mm5, %%mm6\n\t"
- "movq 24(%1, %0, 4), %%mm0\n\t"
- "movq %%mm0, %%mm3\n\t"
- "punpcklbw %%mm6, %%mm0\n\t" /* Y U6 Y V6 Y U6 Y V6*/
- "punpckhbw %%mm6, %%mm3\n\t" /* Y U7 Y V7 Y U7 Y V7*/
- MOVNTQ" %%mm0, 48(%4, %0, 8)\n\t"
- MOVNTQ" %%mm3, 56(%4, %0, 8)\n\t"
-
- : "+r" (x)
- : "r"(yp), "r" (up), "r"(vp), "r"(d)
- :"memory");
- }
-#endif
- for(; x<w; x++)
- {
- const long x2= x<<2;
- d[8*x+0]=yp[x2];
- d[8*x+1]=up[x];
- d[8*x+2]=yp[x2+1];
- d[8*x+3]=vp[x];
- d[8*x+4]=yp[x2+2];
- d[8*x+5]=up[x];
- d[8*x+6]=yp[x2+3];
- d[8*x+7]=vp[x];
- }
+ for (;x<w-7;x+=8)
+ {
+ asm volatile(
+ PREFETCH" 32(%1, %0) \n\t"
+ PREFETCH" 32(%2, %0) \n\t"
+ PREFETCH" 32(%3, %0) \n\t"
+ "movq (%1, %0, 4), %%mm0 \n\t" /* Y0Y1Y2Y3Y4Y5Y6Y7 */
+ "movq (%2, %0), %%mm1 \n\t" /* U0U1U2U3U4U5U6U7 */
+ "movq (%3, %0), %%mm2 \n\t" /* V0V1V2V3V4V5V6V7 */
+ "movq %%mm0, %%mm3 \n\t" /* Y0Y1Y2Y3Y4Y5Y6Y7 */
+ "movq %%mm1, %%mm4 \n\t" /* U0U1U2U3U4U5U6U7 */
+ "movq %%mm2, %%mm5 \n\t" /* V0V1V2V3V4V5V6V7 */
+ "punpcklbw %%mm1, %%mm1 \n\t" /* U0U0 U1U1 U2U2 U3U3 */
+ "punpcklbw %%mm2, %%mm2 \n\t" /* V0V0 V1V1 V2V2 V3V3 */
+ "punpckhbw %%mm4, %%mm4 \n\t" /* U4U4 U5U5 U6U6 U7U7 */
+ "punpckhbw %%mm5, %%mm5 \n\t" /* V4V4 V5V5 V6V6 V7V7 */
+
+ "movq %%mm1, %%mm6 \n\t"
+ "punpcklbw %%mm2, %%mm1 \n\t" /* U0V0 U0V0 U1V1 U1V1*/
+ "punpcklbw %%mm1, %%mm0 \n\t" /* Y0U0 Y1V0 Y2U0 Y3V0*/
+ "punpckhbw %%mm1, %%mm3 \n\t" /* Y4U1 Y5V1 Y6U1 Y7V1*/
+ MOVNTQ" %%mm0, (%4, %0, 8) \n\t"
+ MOVNTQ" %%mm3, 8(%4, %0, 8) \n\t"
+
+ "punpckhbw %%mm2, %%mm6 \n\t" /* U2V2 U2V2 U3V3 U3V3*/
+ "movq 8(%1, %0, 4), %%mm0 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "punpcklbw %%mm6, %%mm0 \n\t" /* Y U2 Y V2 Y U2 Y V2*/
+ "punpckhbw %%mm6, %%mm3 \n\t" /* Y U3 Y V3 Y U3 Y V3*/
+ MOVNTQ" %%mm0, 16(%4, %0, 8) \n\t"
+ MOVNTQ" %%mm3, 24(%4, %0, 8) \n\t"
+
+ "movq %%mm4, %%mm6 \n\t"
+ "movq 16(%1, %0, 4), %%mm0 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "punpcklbw %%mm5, %%mm4 \n\t"
+ "punpcklbw %%mm4, %%mm0 \n\t" /* Y U4 Y V4 Y U4 Y V4*/
+ "punpckhbw %%mm4, %%mm3 \n\t" /* Y U5 Y V5 Y U5 Y V5*/
+ MOVNTQ" %%mm0, 32(%4, %0, 8) \n\t"
+ MOVNTQ" %%mm3, 40(%4, %0, 8) \n\t"
+
+ "punpckhbw %%mm5, %%mm6 \n\t"
+ "movq 24(%1, %0, 4), %%mm0 \n\t"
+ "movq %%mm0, %%mm3 \n\t"
+ "punpcklbw %%mm6, %%mm0 \n\t" /* Y U6 Y V6 Y U6 Y V6*/
+ "punpckhbw %%mm6, %%mm3 \n\t" /* Y U7 Y V7 Y U7 Y V7*/
+ MOVNTQ" %%mm0, 48(%4, %0, 8) \n\t"
+ MOVNTQ" %%mm3, 56(%4, %0, 8) \n\t"
+
+ : "+r" (x)
+ : "r"(yp), "r" (up), "r"(vp), "r"(d)
+ :"memory");
+ }
+#endif
+ for (; x<w; x++)
+ {
+ const long x2 = x<<2;
+ d[8*x+0] = yp[x2];
+ d[8*x+1] = up[x];
+ d[8*x+2] = yp[x2+1];
+ d[8*x+3] = vp[x];
+ d[8*x+4] = yp[x2+2];
+ d[8*x+5] = up[x];
+ d[8*x+6] = yp[x2+3];
+ d[8*x+7] = vp[x];
+ }
}
#ifdef HAVE_MMX
- asm(
- EMMS" \n\t"
- SFENCE" \n\t"
- ::: "memory"
- );
+ asm(
+ EMMS" \n\t"
+ SFENCE" \n\t"
+ ::: "memory"
+ );
#endif
}
static inline void RENAME(rgb2rgb_init)(void){
- rgb15to16= RENAME(rgb15to16);
- rgb15to24= RENAME(rgb15to24);
- rgb15to32= RENAME(rgb15to32);
- rgb16to24= RENAME(rgb16to24);
- rgb16to32= RENAME(rgb16to32);
- rgb16to15= RENAME(rgb16to15);
- rgb24to16= RENAME(rgb24to16);
- rgb24to15= RENAME(rgb24to15);
- rgb24to32= RENAME(rgb24to32);
- rgb32to16= RENAME(rgb32to16);
- rgb32to15= RENAME(rgb32to15);
- rgb32to24= RENAME(rgb32to24);
- rgb24tobgr15= RENAME(rgb24tobgr15);
- rgb24tobgr16= RENAME(rgb24tobgr16);
- rgb24tobgr24= RENAME(rgb24tobgr24);
- rgb32tobgr32= RENAME(rgb32tobgr32);
- rgb32tobgr16= RENAME(rgb32tobgr16);
- rgb32tobgr15= RENAME(rgb32tobgr15);
- yv12toyuy2= RENAME(yv12toyuy2);
- yv12touyvy= RENAME(yv12touyvy);
- yuv422ptoyuy2= RENAME(yuv422ptoyuy2);
- yuy2toyv12= RENAME(yuy2toyv12);
-// uyvytoyv12= RENAME(uyvytoyv12);
-// yvu9toyv12= RENAME(yvu9toyv12);
- planar2x= RENAME(planar2x);
- rgb24toyv12= RENAME(rgb24toyv12);
- interleaveBytes= RENAME(interleaveBytes);
- vu9_to_vu12= RENAME(vu9_to_vu12);
- yvu9_to_yuy2= RENAME(yvu9_to_yuy2);
+ rgb15to16 = RENAME(rgb15to16);
+ rgb15to24 = RENAME(rgb15to24);
+ rgb15to32 = RENAME(rgb15to32);
+ rgb16to24 = RENAME(rgb16to24);
+ rgb16to32 = RENAME(rgb16to32);
+ rgb16to15 = RENAME(rgb16to15);
+ rgb24to16 = RENAME(rgb24to16);
+ rgb24to15 = RENAME(rgb24to15);
+ rgb24to32 = RENAME(rgb24to32);
+ rgb32to16 = RENAME(rgb32to16);
+ rgb32to15 = RENAME(rgb32to15);
+ rgb32to24 = RENAME(rgb32to24);
+ rgb24tobgr15 = RENAME(rgb24tobgr15);
+ rgb24tobgr16 = RENAME(rgb24tobgr16);
+ rgb24tobgr24 = RENAME(rgb24tobgr24);
+ rgb32tobgr32 = RENAME(rgb32tobgr32);
+ rgb32tobgr16 = RENAME(rgb32tobgr16);
+ rgb32tobgr15 = RENAME(rgb32tobgr15);
+ yv12toyuy2 = RENAME(yv12toyuy2);
+ yv12touyvy = RENAME(yv12touyvy);
+ yuv422ptoyuy2 = RENAME(yuv422ptoyuy2);
+ yuy2toyv12 = RENAME(yuy2toyv12);
+// uyvytoyv12 = RENAME(uyvytoyv12);
+// yvu9toyv12 = RENAME(yvu9toyv12);
+ planar2x = RENAME(planar2x);
+ rgb24toyv12 = RENAME(rgb24toyv12);
+ interleaveBytes = RENAME(interleaveBytes);
+ vu9_to_vu12 = RENAME(vu9_to_vu12);
+ yvu9_to_yuy2 = RENAME(yvu9_to_yuy2);
}
diff --git a/libswscale/swscale_altivec_template.c b/libswscale/swscale_altivec_template.c
index d5c0e1f34e..e9b51a808e 100644
--- a/libswscale/swscale_altivec_template.c
+++ b/libswscale/swscale_altivec_template.c
@@ -31,518 +31,514 @@
static inline void
altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW) {
- register int i;
- vector unsigned int altivec_vectorShiftInt19 =
- vec_add(vec_splat_u32(10),vec_splat_u32(9));
- if ((unsigned long)dest % 16) {
- /* badly aligned store, we force store alignement */
- /* and will handle load misalignement on val w/ vec_perm */
- vector unsigned char perm1;
- vector signed int v1;
- for (i = 0 ; (i < dstW) &&
- (((unsigned long)dest + i) % 16) ; i++) {
- int t = val[i] >> 19;
- dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
- }
- perm1 = vec_lvsl(i << 2, val);
- v1 = vec_ld(i << 2, val);
- for ( ; i < (dstW - 15); i+=16) {
- int offset = i << 2;
- vector signed int v2 = vec_ld(offset + 16, val);
- vector signed int v3 = vec_ld(offset + 32, val);
- vector signed int v4 = vec_ld(offset + 48, val);
- vector signed int v5 = vec_ld(offset + 64, val);
- vector signed int v12 = vec_perm(v1,v2,perm1);
- vector signed int v23 = vec_perm(v2,v3,perm1);
- vector signed int v34 = vec_perm(v3,v4,perm1);
- vector signed int v45 = vec_perm(v4,v5,perm1);
-
- vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19);
- vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19);
- vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19);
- vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19);
- vector unsigned short vs1 = vec_packsu(vA, vB);
- vector unsigned short vs2 = vec_packsu(vC, vD);
- vector unsigned char vf = vec_packsu(vs1, vs2);
- vec_st(vf, i, dest);
- v1 = v5;
+ register int i;
+ vector unsigned int altivec_vectorShiftInt19 =
+ vec_add(vec_splat_u32(10),vec_splat_u32(9));
+ if ((unsigned long)dest % 16) {
+ /* badly aligned store, we force store alignement */
+ /* and will handle load misalignement on val w/ vec_perm */
+ vector unsigned char perm1;
+ vector signed int v1;
+ for (i = 0 ; (i < dstW) &&
+ (((unsigned long)dest + i) % 16) ; i++) {
+ int t = val[i] >> 19;
+ dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
+ }
+ perm1 = vec_lvsl(i << 2, val);
+ v1 = vec_ld(i << 2, val);
+ for ( ; i < (dstW - 15); i+=16) {
+ int offset = i << 2;
+ vector signed int v2 = vec_ld(offset + 16, val);
+ vector signed int v3 = vec_ld(offset + 32, val);
+ vector signed int v4 = vec_ld(offset + 48, val);
+ vector signed int v5 = vec_ld(offset + 64, val);
+ vector signed int v12 = vec_perm(v1,v2,perm1);
+ vector signed int v23 = vec_perm(v2,v3,perm1);
+ vector signed int v34 = vec_perm(v3,v4,perm1);
+ vector signed int v45 = vec_perm(v4,v5,perm1);
+
+ vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19);
+ vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19);
+ vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19);
+ vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19);
+ vector unsigned short vs1 = vec_packsu(vA, vB);
+ vector unsigned short vs2 = vec_packsu(vC, vD);
+ vector unsigned char vf = vec_packsu(vs1, vs2);
+ vec_st(vf, i, dest);
+ v1 = v5;
+ }
+ } else { // dest is properly aligned, great
+ for (i = 0; i < (dstW - 15); i+=16) {
+ int offset = i << 2;
+ vector signed int v1 = vec_ld(offset, val);
+ vector signed int v2 = vec_ld(offset + 16, val);
+ vector signed int v3 = vec_ld(offset + 32, val);
+ vector signed int v4 = vec_ld(offset + 48, val);
+ vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19);
+ vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19);
+ vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19);
+ vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19);
+ vector unsigned short vs1 = vec_packsu(v5, v6);
+ vector unsigned short vs2 = vec_packsu(v7, v8);
+ vector unsigned char vf = vec_packsu(vs1, vs2);
+ vec_st(vf, i, dest);
+ }
}
- } else { // dest is properly aligned, great
- for (i = 0; i < (dstW - 15); i+=16) {
- int offset = i << 2;
- vector signed int v1 = vec_ld(offset, val);
- vector signed int v2 = vec_ld(offset + 16, val);
- vector signed int v3 = vec_ld(offset + 32, val);
- vector signed int v4 = vec_ld(offset + 48, val);
- vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19);
- vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19);
- vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19);
- vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19);
- vector unsigned short vs1 = vec_packsu(v5, v6);
- vector unsigned short vs2 = vec_packsu(v7, v8);
- vector unsigned char vf = vec_packsu(vs1, vs2);
- vec_st(vf, i, dest);
+ for ( ; i < dstW ; i++) {
+ int t = val[i] >> 19;
+ dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
}
- }
- for ( ; i < dstW ; i++) {
- int t = val[i] >> 19;
- dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
- }
}
static inline void
yuv2yuvX_altivec_real(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
- int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
- uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
+ int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
+ uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
{
- const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
- register int i, j;
- {
- int __attribute__ ((aligned (16))) val[dstW];
+ const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
+ register int i, j;
+ {
+ int __attribute__ ((aligned (16))) val[dstW];
- for (i = 0; i < (dstW -7); i+=4) {
- vec_st(vini, i << 2, val);
- }
- for (; i < dstW; i++) {
- val[i] = (1 << 18);
- }
+ for (i = 0; i < (dstW -7); i+=4) {
+ vec_st(vini, i << 2, val);
+ }
+ for (; i < dstW; i++) {
+ val[i] = (1 << 18);
+ }
- for (j = 0; j < lumFilterSize; j++) {
- vector signed short l1, vLumFilter = vec_ld(j << 1, lumFilter);
- vector unsigned char perm, perm0 = vec_lvsl(j << 1, lumFilter);
- vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0);
- vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter
+ for (j = 0; j < lumFilterSize; j++) {
+ vector signed short l1, vLumFilter = vec_ld(j << 1, lumFilter);
+ vector unsigned char perm, perm0 = vec_lvsl(j << 1, lumFilter);
+ vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0);
+ vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter
- perm = vec_lvsl(0, lumSrc[j]);
- l1 = vec_ld(0, lumSrc[j]);
+ perm = vec_lvsl(0, lumSrc[j]);
+ l1 = vec_ld(0, lumSrc[j]);
- for (i = 0; i < (dstW - 7); i+=8) {
- int offset = i << 2;
- vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]);
+ for (i = 0; i < (dstW - 7); i+=8) {
+ int offset = i << 2;
+ vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]);
- vector signed int v1 = vec_ld(offset, val);
- vector signed int v2 = vec_ld(offset + 16, val);
+ vector signed int v1 = vec_ld(offset, val);
+ vector signed int v2 = vec_ld(offset + 16, val);
- vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7]
+ vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7]
- vector signed int i1 = vec_mule(vLumFilter, ls);
- vector signed int i2 = vec_mulo(vLumFilter, ls);
+ vector signed int i1 = vec_mule(vLumFilter, ls);
+ vector signed int i2 = vec_mulo(vLumFilter, ls);
- vector signed int vf1 = vec_mergeh(i1, i2);
- vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j]
+ vector signed int vf1 = vec_mergeh(i1, i2);
+ vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j]
- vector signed int vo1 = vec_add(v1, vf1);
- vector signed int vo2 = vec_add(v2, vf2);
+ vector signed int vo1 = vec_add(v1, vf1);
+ vector signed int vo2 = vec_add(v2, vf2);
- vec_st(vo1, offset, val);
- vec_st(vo2, offset + 16, val);
+ vec_st(vo1, offset, val);
+ vec_st(vo2, offset + 16, val);
- l1 = l2;
- }
- for ( ; i < dstW; i++) {
- val[i] += lumSrc[j][i] * lumFilter[j];
- }
- }
- altivec_packIntArrayToCharArray(val,dest,dstW);
- }
- if (uDest != 0) {
- int __attribute__ ((aligned (16))) u[chrDstW];
- int __attribute__ ((aligned (16))) v[chrDstW];
-
- for (i = 0; i < (chrDstW -7); i+=4) {
- vec_st(vini, i << 2, u);
- vec_st(vini, i << 2, v);
- }
- for (; i < chrDstW; i++) {
- u[i] = (1 << 18);
- v[i] = (1 << 18);
+ l1 = l2;
+ }
+ for ( ; i < dstW; i++) {
+ val[i] += lumSrc[j][i] * lumFilter[j];
+ }
+ }
+ altivec_packIntArrayToCharArray(val,dest,dstW);
}
+ if (uDest != 0) {
+ int __attribute__ ((aligned (16))) u[chrDstW];
+ int __attribute__ ((aligned (16))) v[chrDstW];
+
+ for (i = 0; i < (chrDstW -7); i+=4) {
+ vec_st(vini, i << 2, u);
+ vec_st(vini, i << 2, v);
+ }
+ for (; i < chrDstW; i++) {
+ u[i] = (1 << 18);
+ v[i] = (1 << 18);
+ }
- for (j = 0; j < chrFilterSize; j++) {
- vector signed short l1, l1_V, vChrFilter = vec_ld(j << 1, chrFilter);
- vector unsigned char perm, perm0 = vec_lvsl(j << 1, chrFilter);
- vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
- vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter
-
- perm = vec_lvsl(0, chrSrc[j]);
- l1 = vec_ld(0, chrSrc[j]);
- l1_V = vec_ld(2048 << 1, chrSrc[j]);
-
- for (i = 0; i < (chrDstW - 7); i+=8) {
- int offset = i << 2;
- vector signed short l2 = vec_ld((i << 1) + 16, chrSrc[j]);
- vector signed short l2_V = vec_ld(((i + 2048) << 1) + 16, chrSrc[j]);
-
- vector signed int v1 = vec_ld(offset, u);
- vector signed int v2 = vec_ld(offset + 16, u);
- vector signed int v1_V = vec_ld(offset, v);
- vector signed int v2_V = vec_ld(offset + 16, v);
-
- vector signed short ls = vec_perm(l1, l2, perm); // chrSrc[j][i] ... chrSrc[j][i+7]
- vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrSrc[j][i+2048] ... chrSrc[j][i+2055]
-
- vector signed int i1 = vec_mule(vChrFilter, ls);
- vector signed int i2 = vec_mulo(vChrFilter, ls);
- vector signed int i1_V = vec_mule(vChrFilter, ls_V);
- vector signed int i2_V = vec_mulo(vChrFilter, ls_V);
-
- vector signed int vf1 = vec_mergeh(i1, i2);
- vector signed int vf2 = vec_mergel(i1, i2); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
- vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
- vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
-
- vector signed int vo1 = vec_add(v1, vf1);
- vector signed int vo2 = vec_add(v2, vf2);
- vector signed int vo1_V = vec_add(v1_V, vf1_V);
- vector signed int vo2_V = vec_add(v2_V, vf2_V);
-
- vec_st(vo1, offset, u);
- vec_st(vo2, offset + 16, u);
- vec_st(vo1_V, offset, v);
- vec_st(vo2_V, offset + 16, v);
-
- l1 = l2;
- l1_V = l2_V;
- }
- for ( ; i < chrDstW; i++) {
- u[i] += chrSrc[j][i] * chrFilter[j];
- v[i] += chrSrc[j][i + 2048] * chrFilter[j];
- }
+ for (j = 0; j < chrFilterSize; j++) {
+ vector signed short l1, l1_V, vChrFilter = vec_ld(j << 1, chrFilter);
+ vector unsigned char perm, perm0 = vec_lvsl(j << 1, chrFilter);
+ vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
+ vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter
+
+ perm = vec_lvsl(0, chrSrc[j]);
+ l1 = vec_ld(0, chrSrc[j]);
+ l1_V = vec_ld(2048 << 1, chrSrc[j]);
+
+ for (i = 0; i < (chrDstW - 7); i+=8) {
+ int offset = i << 2;
+ vector signed short l2 = vec_ld((i << 1) + 16, chrSrc[j]);
+ vector signed short l2_V = vec_ld(((i + 2048) << 1) + 16, chrSrc[j]);
+
+ vector signed int v1 = vec_ld(offset, u);
+ vector signed int v2 = vec_ld(offset + 16, u);
+ vector signed int v1_V = vec_ld(offset, v);
+ vector signed int v2_V = vec_ld(offset + 16, v);
+
+ vector signed short ls = vec_perm(l1, l2, perm); // chrSrc[j][i] ... chrSrc[j][i+7]
+ vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrSrc[j][i+2048] ... chrSrc[j][i+2055]
+
+ vector signed int i1 = vec_mule(vChrFilter, ls);
+ vector signed int i2 = vec_mulo(vChrFilter, ls);
+ vector signed int i1_V = vec_mule(vChrFilter, ls_V);
+ vector signed int i2_V = vec_mulo(vChrFilter, ls_V);
+
+ vector signed int vf1 = vec_mergeh(i1, i2);
+ vector signed int vf2 = vec_mergel(i1, i2); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
+ vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
+ vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
+
+ vector signed int vo1 = vec_add(v1, vf1);
+ vector signed int vo2 = vec_add(v2, vf2);
+ vector signed int vo1_V = vec_add(v1_V, vf1_V);
+ vector signed int vo2_V = vec_add(v2_V, vf2_V);
+
+ vec_st(vo1, offset, u);
+ vec_st(vo2, offset + 16, u);
+ vec_st(vo1_V, offset, v);
+ vec_st(vo2_V, offset + 16, v);
+
+ l1 = l2;
+ l1_V = l2_V;
+ }
+ for ( ; i < chrDstW; i++) {
+ u[i] += chrSrc[j][i] * chrFilter[j];
+ v[i] += chrSrc[j][i + 2048] * chrFilter[j];
+ }
+ }
+ altivec_packIntArrayToCharArray(u,uDest,chrDstW);
+ altivec_packIntArrayToCharArray(v,vDest,chrDstW);
}
- altivec_packIntArrayToCharArray(u,uDest,chrDstW);
- altivec_packIntArrayToCharArray(v,vDest,chrDstW);
- }
}
static inline void hScale_altivec_real(int16_t *dst, int dstW, uint8_t *src, int srcW, int xInc, int16_t *filter, int16_t *filterPos, int filterSize) {
- register int i;
- int __attribute__ ((aligned (16))) tempo[4];
-
- if (filterSize % 4) {
- for(i=0; i<dstW; i++) {
- register int j;
- register int srcPos = filterPos[i];
- register int val = 0;
- for(j=0; j<filterSize; j++) {
- val += ((int)src[srcPos + j])*filter[filterSize*i + j];
- }
- dst[i] = av_clip(val>>7, 0, (1<<15)-1);
+ register int i;
+ int __attribute__ ((aligned (16))) tempo[4];
+
+ if (filterSize % 4) {
+ for (i=0; i<dstW; i++) {
+ register int j;
+ register int srcPos = filterPos[i];
+ register int val = 0;
+ for (j=0; j<filterSize; j++) {
+ val += ((int)src[srcPos + j])*filter[filterSize*i + j];
+ }
+ dst[i] = av_clip(val>>7, 0, (1<<15)-1);
+ }
}
- }
- else
- switch (filterSize) {
- case 4:
+ else
+ switch (filterSize) {
+ case 4:
{
- for(i=0; i<dstW; i++) {
- register int srcPos = filterPos[i];
-
- vector unsigned char src_v0 = vec_ld(srcPos, src);
- vector unsigned char src_v1, src_vF;
- vector signed short src_v, filter_v;
- vector signed int val_vEven, val_s;
- if ((((int)src + srcPos)% 16) > 12) {
- src_v1 = vec_ld(srcPos + 16, src);
- }
- src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
-
- src_v = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
- // now put our elements in the even slots
- src_v = vec_mergeh(src_v, (vector signed short)vzero);
-
- filter_v = vec_ld(i << 3, filter);
+ for (i=0; i<dstW; i++) {
+ register int srcPos = filterPos[i];
+
+ vector unsigned char src_v0 = vec_ld(srcPos, src);
+ vector unsigned char src_v1, src_vF;
+ vector signed short src_v, filter_v;
+ vector signed int val_vEven, val_s;
+ if ((((int)src + srcPos)% 16) > 12) {
+ src_v1 = vec_ld(srcPos + 16, src);
+ }
+ src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+ src_v = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+ // now put our elements in the even slots
+ src_v = vec_mergeh(src_v, (vector signed short)vzero);
+
+ filter_v = vec_ld(i << 3, filter);
// the 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2)
// the neat trick : we only care for half the elements,
// high or low depending on (i<<3)%16 (it's 0 or 8 here),
// and we're going to use vec_mule, so we chose
// carefully how to "unpack" the elements into the even slots
- if ((i << 3) % 16)
- filter_v = vec_mergel(filter_v,(vector signed short)vzero);
- else
- filter_v = vec_mergeh(filter_v,(vector signed short)vzero);
-
- val_vEven = vec_mule(src_v, filter_v);
- val_s = vec_sums(val_vEven, vzero);
- vec_st(val_s, 0, tempo);
- dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
- }
+ if ((i << 3) % 16)
+ filter_v = vec_mergel(filter_v,(vector signed short)vzero);
+ else
+ filter_v = vec_mergeh(filter_v,(vector signed short)vzero);
+
+ val_vEven = vec_mule(src_v, filter_v);
+ val_s = vec_sums(val_vEven, vzero);
+ vec_st(val_s, 0, tempo);
+ dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
+ }
}
break;
- case 8:
+ case 8:
{
- for(i=0; i<dstW; i++) {
- register int srcPos = filterPos[i];
-
- vector unsigned char src_v0 = vec_ld(srcPos, src);
- vector unsigned char src_v1, src_vF;
- vector signed short src_v, filter_v;
- vector signed int val_v, val_s;
- if ((((int)src + srcPos)% 16) > 8) {
- src_v1 = vec_ld(srcPos + 16, src);
- }
- src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
-
- src_v = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
- filter_v = vec_ld(i << 4, filter);
+ for (i=0; i<dstW; i++) {
+ register int srcPos = filterPos[i];
+
+ vector unsigned char src_v0 = vec_ld(srcPos, src);
+ vector unsigned char src_v1, src_vF;
+ vector signed short src_v, filter_v;
+ vector signed int val_v, val_s;
+ if ((((int)src + srcPos)% 16) > 8) {
+ src_v1 = vec_ld(srcPos + 16, src);
+ }
+ src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+
+ src_v = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+ filter_v = vec_ld(i << 4, filter);
// the 4 above is 3 (filterSize == 8) + 1 (sizeof(short) == 2)
- val_v = vec_msums(src_v, filter_v, (vector signed int)vzero);
- val_s = vec_sums(val_v, vzero);
- vec_st(val_s, 0, tempo);
- dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
- }
+ val_v = vec_msums(src_v, filter_v, (vector signed int)vzero);
+ val_s = vec_sums(val_v, vzero);
+ vec_st(val_s, 0, tempo);
+ dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
+ }
}
break;
- case 16:
+ case 16:
{
- for(i=0; i<dstW; i++) {
- register int srcPos = filterPos[i];
+ for (i=0; i<dstW; i++) {
+ register int srcPos = filterPos[i];
- vector unsigned char src_v0 = vec_ld(srcPos, src);
- vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
- vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
+ vector unsigned char src_v0 = vec_ld(srcPos, src);
+ vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
+ vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
- vector signed short src_vA = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
- vector signed short src_vB = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
+ vector signed short src_vA = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+ vector signed short src_vB = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
- vector signed short filter_v0 = vec_ld(i << 5, filter);
- vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
- // the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2)
+ vector signed short filter_v0 = vec_ld(i << 5, filter);
+ vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
+ // the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2)
- vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero);
- vector signed int val_v = vec_msums(src_vB, filter_v1, val_acc);
+ vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero);
+ vector signed int val_v = vec_msums(src_vB, filter_v1, val_acc);
- vector signed int val_s = vec_sums(val_v, vzero);
+ vector signed int val_s = vec_sums(val_v, vzero);
- vec_st(val_s, 0, tempo);
- dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
- }
+ vec_st(val_s, 0, tempo);
+ dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
+ }
}
break;
- default:
+ default:
{
- for(i=0; i<dstW; i++) {
- register int j;
- register int srcPos = filterPos[i];
+ for (i=0; i<dstW; i++) {
+ register int j;
+ register int srcPos = filterPos[i];
vector signed int val_s, val_v = (vector signed int)vzero;
- vector signed short filter_v0R = vec_ld(i * 2 * filterSize, filter);
+ vector signed short filter_v0R = vec_ld(i * 2 * filterSize, filter);
vector unsigned char permF = vec_lvsl((i * 2 * filterSize), filter);
vector unsigned char src_v0 = vec_ld(srcPos, src);
vector unsigned char permS = vec_lvsl(srcPos, src);
for (j = 0 ; j < filterSize - 15; j += 16) {
- vector unsigned char src_v1 = vec_ld(srcPos + j + 16, src);
- vector unsigned char src_vF = vec_perm(src_v0, src_v1, permS);
+ vector unsigned char src_v1 = vec_ld(srcPos + j + 16, src);
+ vector unsigned char src_vF = vec_perm(src_v0, src_v1, permS);
- vector signed short src_vA = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
- vector signed short src_vB = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
+ vector signed short src_vA = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+ vector signed short src_vB = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
- vector signed short filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
- vector signed short filter_v2R = vec_ld((i * 2 * filterSize) + (j * 2) + 32, filter);
- vector signed short filter_v0 = vec_perm(filter_v0R, filter_v1R, permF);
- vector signed short filter_v1 = vec_perm(filter_v1R, filter_v2R, permF);
+ vector signed short filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
+ vector signed short filter_v2R = vec_ld((i * 2 * filterSize) + (j * 2) + 32, filter);
+ vector signed short filter_v0 = vec_perm(filter_v0R, filter_v1R, permF);
+ vector signed short filter_v1 = vec_perm(filter_v1R, filter_v2R, permF);
- vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v);
- val_v = vec_msums(src_vB, filter_v1, val_acc);
+ vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v);
+ val_v = vec_msums(src_vB, filter_v1, val_acc);
- filter_v0R = filter_v2R;
- src_v0 = src_v1;
+ filter_v0R = filter_v2R;
+ src_v0 = src_v1;
}
if (j < (filterSize-7)) {
- // loading src_v0 is useless, it's already done above
- //vector unsigned char src_v0 = vec_ld(srcPos + j, src);
- vector unsigned char src_v1, src_vF;
- vector signed short src_v, filter_v1R, filter_v;
- if ((((int)src + srcPos)% 16) > 8) {
- src_v1 = vec_ld(srcPos + j + 16, src);
- }
- src_vF = vec_perm(src_v0, src_v1, permS);
-
- src_v = // vec_unpackh sign-extends...
- (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
- // loading filter_v0R is useless, it's already done above
- //vector signed short filter_v0R = vec_ld((i * 2 * filterSize) + j, filter);
- filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
- filter_v = vec_perm(filter_v0R, filter_v1R, permF);
-
- val_v = vec_msums(src_v, filter_v, val_v);
+ // loading src_v0 is useless, it's already done above
+ //vector unsigned char src_v0 = vec_ld(srcPos + j, src);
+ vector unsigned char src_v1, src_vF;
+ vector signed short src_v, filter_v1R, filter_v;
+ if ((((int)src + srcPos)% 16) > 8) {
+ src_v1 = vec_ld(srcPos + j + 16, src);
+ }
+ src_vF = vec_perm(src_v0, src_v1, permS);
+
+ src_v = // vec_unpackh sign-extends...
+ (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
+ // loading filter_v0R is useless, it's already done above
+ //vector signed short filter_v0R = vec_ld((i * 2 * filterSize) + j, filter);
+ filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
+ filter_v = vec_perm(filter_v0R, filter_v1R, permF);
+
+ val_v = vec_msums(src_v, filter_v, val_v);
}
val_s = vec_sums(val_v, vzero);
vec_st(val_s, 0, tempo);
dst[i] = av_clip(tempo[3]>>7, 0, (1<<15)-1);
- }
+ }
}
- }
+ }
}
static inline int yv12toyuy2_unscaled_altivec(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
- int srcSliceH, uint8_t* dstParam[], int dstStride_a[]) {
- uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
- // yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] );
- uint8_t *ysrc = src[0];
- uint8_t *usrc = src[1];
- uint8_t *vsrc = src[2];
- const int width = c->srcW;
- const int height = srcSliceH;
- const int lumStride = srcStride[0];
- const int chromStride = srcStride[1];
- const int dstStride = dstStride_a[0];
- const vector unsigned char yperm = vec_lvsl(0, ysrc);
- const int vertLumPerChroma = 2;
- register unsigned int y;
-
- if(width&15){
- yv12toyuy2( ysrc, usrc, vsrc, dst,c->srcW,srcSliceH, lumStride, chromStride, dstStride);
- return srcSliceH;
- }
-
- /* this code assume:
-
- 1) dst is 16 bytes-aligned
- 2) dstStride is a multiple of 16
- 3) width is a multiple of 16
- 4) lum&chrom stride are multiple of 8
- */
+ int srcSliceH, uint8_t* dstParam[], int dstStride_a[]) {
+ uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
+ // yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] );
+ uint8_t *ysrc = src[0];
+ uint8_t *usrc = src[1];
+ uint8_t *vsrc = src[2];
+ const int width = c->srcW;
+ const int height = srcSliceH;
+ const int lumStride = srcStride[0];
+ const int chromStride = srcStride[1];
+ const int dstStride = dstStride_a[0];
+ const vector unsigned char yperm = vec_lvsl(0, ysrc);
+ const int vertLumPerChroma = 2;
+ register unsigned int y;
+
+ if (width&15) {
+ yv12toyuy2( ysrc, usrc, vsrc, dst,c->srcW,srcSliceH, lumStride, chromStride, dstStride);
+ return srcSliceH;
+ }
- for(y=0; y<height; y++)
- {
- int i;
- for (i = 0; i < width - 31; i+= 32) {
- const unsigned int j = i >> 1;
- vector unsigned char v_yA = vec_ld(i, ysrc);
- vector unsigned char v_yB = vec_ld(i + 16, ysrc);
- vector unsigned char v_yC = vec_ld(i + 32, ysrc);
- vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
- vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
- vector unsigned char v_uA = vec_ld(j, usrc);
- vector unsigned char v_uB = vec_ld(j + 16, usrc);
- vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
- vector unsigned char v_vA = vec_ld(j, vsrc);
- vector unsigned char v_vB = vec_ld(j + 16, vsrc);
- vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
- vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
- vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
- vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
- vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
- vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
- vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
- vec_st(v_yuy2_0, (i << 1), dst);
- vec_st(v_yuy2_1, (i << 1) + 16, dst);
- vec_st(v_yuy2_2, (i << 1) + 32, dst);
- vec_st(v_yuy2_3, (i << 1) + 48, dst);
- }
- if (i < width) {
- const unsigned int j = i >> 1;
- vector unsigned char v_y1 = vec_ld(i, ysrc);
- vector unsigned char v_u = vec_ld(j, usrc);
- vector unsigned char v_v = vec_ld(j, vsrc);
- vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
- vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
- vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
- vec_st(v_yuy2_0, (i << 1), dst);
- vec_st(v_yuy2_1, (i << 1) + 16, dst);
- }
- if((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
- {
- usrc += chromStride;
- vsrc += chromStride;
- }
- ysrc += lumStride;
- dst += dstStride;
+ /* this code assume:
+
+ 1) dst is 16 bytes-aligned
+ 2) dstStride is a multiple of 16
+ 3) width is a multiple of 16
+ 4) lum&chrom stride are multiple of 8
+ */
+
+ for (y=0; y<height; y++) {
+ int i;
+ for (i = 0; i < width - 31; i+= 32) {
+ const unsigned int j = i >> 1;
+ vector unsigned char v_yA = vec_ld(i, ysrc);
+ vector unsigned char v_yB = vec_ld(i + 16, ysrc);
+ vector unsigned char v_yC = vec_ld(i + 32, ysrc);
+ vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
+ vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
+ vector unsigned char v_uA = vec_ld(j, usrc);
+ vector unsigned char v_uB = vec_ld(j + 16, usrc);
+ vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
+ vector unsigned char v_vA = vec_ld(j, vsrc);
+ vector unsigned char v_vB = vec_ld(j + 16, vsrc);
+ vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
+ vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+ vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
+ vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
+ vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
+ vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
+ vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
+ vec_st(v_yuy2_0, (i << 1), dst);
+ vec_st(v_yuy2_1, (i << 1) + 16, dst);
+ vec_st(v_yuy2_2, (i << 1) + 32, dst);
+ vec_st(v_yuy2_3, (i << 1) + 48, dst);
+ }
+ if (i < width) {
+ const unsigned int j = i >> 1;
+ vector unsigned char v_y1 = vec_ld(i, ysrc);
+ vector unsigned char v_u = vec_ld(j, usrc);
+ vector unsigned char v_v = vec_ld(j, vsrc);
+ vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+ vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
+ vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
+ vec_st(v_yuy2_0, (i << 1), dst);
+ vec_st(v_yuy2_1, (i << 1) + 16, dst);
+ }
+ if ( (y&(vertLumPerChroma-1))==(vertLumPerChroma-1) ) {
+ usrc += chromStride;
+ vsrc += chromStride;
+ }
+ ysrc += lumStride;
+ dst += dstStride;
}
- return srcSliceH;
+ return srcSliceH;
}
static inline int yv12touyvy_unscaled_altivec(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
- int srcSliceH, uint8_t* dstParam[], int dstStride_a[]) {
- uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
- // yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] );
- uint8_t *ysrc = src[0];
- uint8_t *usrc = src[1];
- uint8_t *vsrc = src[2];
- const int width = c->srcW;
- const int height = srcSliceH;
- const int lumStride = srcStride[0];
- const int chromStride = srcStride[1];
- const int dstStride = dstStride_a[0];
- const int vertLumPerChroma = 2;
- const vector unsigned char yperm = vec_lvsl(0, ysrc);
- register unsigned int y;
-
- if(width&15){
- yv12touyvy( ysrc, usrc, vsrc, dst,c->srcW,srcSliceH, lumStride, chromStride, dstStride);
- return srcSliceH;
- }
-
- /* this code assume:
-
- 1) dst is 16 bytes-aligned
- 2) dstStride is a multiple of 16
- 3) width is a multiple of 16
- 4) lum&chrom stride are multiple of 8
- */
+ int srcSliceH, uint8_t* dstParam[], int dstStride_a[]) {
+ uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
+ // yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] );
+ uint8_t *ysrc = src[0];
+ uint8_t *usrc = src[1];
+ uint8_t *vsrc = src[2];
+ const int width = c->srcW;
+ const int height = srcSliceH;
+ const int lumStride = srcStride[0];
+ const int chromStride = srcStride[1];
+ const int dstStride = dstStride_a[0];
+ const int vertLumPerChroma = 2;
+ const vector unsigned char yperm = vec_lvsl(0, ysrc);
+ register unsigned int y;
+
+ if (width&15) {
+ yv12touyvy( ysrc, usrc, vsrc, dst,c->srcW,srcSliceH, lumStride, chromStride, dstStride);
+ return srcSliceH;
+ }
- for(y=0; y<height; y++)
- {
- int i;
- for (i = 0; i < width - 31; i+= 32) {
- const unsigned int j = i >> 1;
- vector unsigned char v_yA = vec_ld(i, ysrc);
- vector unsigned char v_yB = vec_ld(i + 16, ysrc);
- vector unsigned char v_yC = vec_ld(i + 32, ysrc);
- vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
- vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
- vector unsigned char v_uA = vec_ld(j, usrc);
- vector unsigned char v_uB = vec_ld(j + 16, usrc);
- vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
- vector unsigned char v_vA = vec_ld(j, vsrc);
- vector unsigned char v_vB = vec_ld(j + 16, vsrc);
- vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
- vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
- vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
- vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
- vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
- vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
- vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
- vec_st(v_uyvy_0, (i << 1), dst);
- vec_st(v_uyvy_1, (i << 1) + 16, dst);
- vec_st(v_uyvy_2, (i << 1) + 32, dst);
- vec_st(v_uyvy_3, (i << 1) + 48, dst);
- }
- if (i < width) {
- const unsigned int j = i >> 1;
- vector unsigned char v_y1 = vec_ld(i, ysrc);
- vector unsigned char v_u = vec_ld(j, usrc);
- vector unsigned char v_v = vec_ld(j, vsrc);
- vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
- vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
- vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
- vec_st(v_uyvy_0, (i << 1), dst);
- vec_st(v_uyvy_1, (i << 1) + 16, dst);
- }
- if((y&(vertLumPerChroma-1))==(vertLumPerChroma-1) )
- {
- usrc += chromStride;
- vsrc += chromStride;
- }
- ysrc += lumStride;
- dst += dstStride;
+ /* this code assume:
+
+ 1) dst is 16 bytes-aligned
+ 2) dstStride is a multiple of 16
+ 3) width is a multiple of 16
+ 4) lum&chrom stride are multiple of 8
+ */
+
+ for (y=0; y<height; y++) {
+ int i;
+ for (i = 0; i < width - 31; i+= 32) {
+ const unsigned int j = i >> 1;
+ vector unsigned char v_yA = vec_ld(i, ysrc);
+ vector unsigned char v_yB = vec_ld(i + 16, ysrc);
+ vector unsigned char v_yC = vec_ld(i + 32, ysrc);
+ vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
+ vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
+ vector unsigned char v_uA = vec_ld(j, usrc);
+ vector unsigned char v_uB = vec_ld(j + 16, usrc);
+ vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
+ vector unsigned char v_vA = vec_ld(j, vsrc);
+ vector unsigned char v_vB = vec_ld(j + 16, vsrc);
+ vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
+ vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+ vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
+ vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
+ vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
+ vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
+ vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
+ vec_st(v_uyvy_0, (i << 1), dst);
+ vec_st(v_uyvy_1, (i << 1) + 16, dst);
+ vec_st(v_uyvy_2, (i << 1) + 32, dst);
+ vec_st(v_uyvy_3, (i << 1) + 48, dst);
+ }
+ if (i < width) {
+ const unsigned int j = i >> 1;
+ vector unsigned char v_y1 = vec_ld(i, ysrc);
+ vector unsigned char v_u = vec_ld(j, usrc);
+ vector unsigned char v_v = vec_ld(j, vsrc);
+ vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
+ vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
+ vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
+ vec_st(v_uyvy_0, (i << 1), dst);
+ vec_st(v_uyvy_1, (i << 1) + 16, dst);
+ }
+ if ( (y&(vertLumPerChroma-1))==(vertLumPerChroma-1) ) {
+ usrc += chromStride;
+ vsrc += chromStride;
+ }
+ ysrc += lumStride;
+ dst += dstStride;
}
- return srcSliceH;
+ return srcSliceH;
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-22 16:54:29 +0000