--- a/rgb2rgb_template.c
+++ b/rgb2rgb_template.c
@@ -77,16 +77,16 @@
     const uint8_t *s = src;
     const uint8_t *end;
     #if HAVE_MMX
-        const uint8_t *mm_end;
+    const uint8_t *mm_end;
     #endif
     end = s + src_size;
     #if HAVE_MMX
-        __asm__ volatile(PREFETCH"    %0"::"m"(*s):"memory");
-        mm_end = end - 23;
-        __asm__ volatile("movq        %0, %%mm7"::"m"(mask32a):"memory");
-        while (s < mm_end)
-        {
-            __asm__ volatile(
+    __asm__ volatile(PREFETCH"    %0"::"m"(*s):"memory");
+    mm_end = end - 23;
+    __asm__ volatile("movq        %0, %%mm7"::"m"(mask32a):"memory");
+    while (s < mm_end)
+    {
+        __asm__ volatile(
             PREFETCH"    32%1           \n\t"
             "movd          %1, %%mm0    \n\t"
             "punpckldq    3%1, %%mm0    \n\t"
@@ -107,11 +107,11 @@
             :"=m"(*dest)
             :"m"(*s)
             :"memory");
-            dest += 32;
-            s += 24;
-        }
-        __asm__ volatile(SFENCE:::"memory");
-        __asm__ volatile(EMMS:::"memory");
+        dest += 32;
+        s += 24;
+    }
+    __asm__ volatile(SFENCE:::"memory");
+    __asm__ volatile(EMMS:::"memory");
     #endif
     while (s < end)
     {
@@ -146,53 +146,53 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movq          %1, %%mm0    \n\t"
-        "movq         8%1, %%mm1    \n\t"
-        "movq        16%1, %%mm4    \n\t"
-        "movq        24%1, %%mm5    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm1, %%mm3    \n\t"
-        "movq       %%mm4, %%mm6    \n\t"
-        "movq       %%mm5, %%mm7    \n\t"
-        "psrlq         $8, %%mm2    \n\t"
-        "psrlq         $8, %%mm3    \n\t"
-        "psrlq         $8, %%mm6    \n\t"
-        "psrlq         $8, %%mm7    \n\t"
-        "pand          %2, %%mm0    \n\t"
-        "pand          %2, %%mm1    \n\t"
-        "pand          %2, %%mm4    \n\t"
-        "pand          %2, %%mm5    \n\t"
-        "pand          %3, %%mm2    \n\t"
-        "pand          %3, %%mm3    \n\t"
-        "pand          %3, %%mm6    \n\t"
-        "pand          %3, %%mm7    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm3, %%mm1    \n\t"
-        "por        %%mm6, %%mm4    \n\t"
-        "por        %%mm7, %%mm5    \n\t"
-
-        "movq       %%mm1, %%mm2    \n\t"
-        "movq       %%mm4, %%mm3    \n\t"
-        "psllq        $48, %%mm2    \n\t"
-        "psllq        $32, %%mm3    \n\t"
-        "pand          %4, %%mm2    \n\t"
-        "pand          %5, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "psrlq        $16, %%mm1    \n\t"
-        "psrlq        $32, %%mm4    \n\t"
-        "psllq        $16, %%mm5    \n\t"
-        "por        %%mm3, %%mm1    \n\t"
-        "pand          %6, %%mm5    \n\t"
-        "por        %%mm5, %%mm4    \n\t"
-
-        MOVNTQ"     %%mm0,   %0     \n\t"
-        MOVNTQ"     %%mm1,  8%0     \n\t"
-        MOVNTQ"     %%mm4, 16%0"
-        :"=m"(*dest)
-        :"m"(*s),"m"(mask24l),
-         "m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
-        :"memory");
+            PREFETCH"    32%1           \n\t"
+            "movq          %1, %%mm0    \n\t"
+            "movq         8%1, %%mm1    \n\t"
+            "movq        16%1, %%mm4    \n\t"
+            "movq        24%1, %%mm5    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm1, %%mm3    \n\t"
+            "movq       %%mm4, %%mm6    \n\t"
+            "movq       %%mm5, %%mm7    \n\t"
+            "psrlq         $8, %%mm2    \n\t"
+            "psrlq         $8, %%mm3    \n\t"
+            "psrlq         $8, %%mm6    \n\t"
+            "psrlq         $8, %%mm7    \n\t"
+            "pand          %2, %%mm0    \n\t"
+            "pand          %2, %%mm1    \n\t"
+            "pand          %2, %%mm4    \n\t"
+            "pand          %2, %%mm5    \n\t"
+            "pand          %3, %%mm2    \n\t"
+            "pand          %3, %%mm3    \n\t"
+            "pand          %3, %%mm6    \n\t"
+            "pand          %3, %%mm7    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm3, %%mm1    \n\t"
+            "por        %%mm6, %%mm4    \n\t"
+            "por        %%mm7, %%mm5    \n\t"
+
+            "movq       %%mm1, %%mm2    \n\t"
+            "movq       %%mm4, %%mm3    \n\t"
+            "psllq        $48, %%mm2    \n\t"
+            "psllq        $32, %%mm3    \n\t"
+            "pand          %4, %%mm2    \n\t"
+            "pand          %5, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "psrlq        $16, %%mm1    \n\t"
+            "psrlq        $32, %%mm4    \n\t"
+            "psllq        $16, %%mm5    \n\t"
+            "por        %%mm3, %%mm1    \n\t"
+            "pand          %6, %%mm5    \n\t"
+            "por        %%mm5, %%mm4    \n\t"
+
+            MOVNTQ"     %%mm0,   %0     \n\t"
+            MOVNTQ"     %%mm1,  8%0     \n\t"
+            MOVNTQ"     %%mm4, 16%0"
+            :"=m"(*dest)
+            :"m"(*s),"m"(mask24l),
+            "m"(mask24h),"m"(mask24hh),"m"(mask24hhh),"m"(mask24hhhh)
+            :"memory");
         dest += 24;
         s += 32;
     }
@@ -237,19 +237,19 @@
     while (s<mm_end)
     {
         __asm__ volatile(
-        PREFETCH"  32%1         \n\t"
-        "movq        %1, %%mm0  \n\t"
-        "movq       8%1, %%mm2  \n\t"
-        "movq     %%mm0, %%mm1  \n\t"
-        "movq     %%mm2, %%mm3  \n\t"
-        "pand     %%mm4, %%mm0  \n\t"
-        "pand     %%mm4, %%mm2  \n\t"
-        "paddw    %%mm1, %%mm0  \n\t"
-        "paddw    %%mm3, %%mm2  \n\t"
-        MOVNTQ"   %%mm0,  %0    \n\t"
-        MOVNTQ"   %%mm2, 8%0"
-        :"=m"(*d)
-        :"m"(*s)
+            PREFETCH"  32%1         \n\t"
+            "movq        %1, %%mm0  \n\t"
+            "movq       8%1, %%mm2  \n\t"
+            "movq     %%mm0, %%mm1  \n\t"
+            "movq     %%mm2, %%mm3  \n\t"
+            "pand     %%mm4, %%mm0  \n\t"
+            "pand     %%mm4, %%mm2  \n\t"
+            "paddw    %%mm1, %%mm0  \n\t"
+            "paddw    %%mm3, %%mm2  \n\t"
+            MOVNTQ"   %%mm0,  %0    \n\t"
+            MOVNTQ"   %%mm2, 8%0"
+            :"=m"(*d)
+            :"m"(*s)
         );
         d+=16;
         s+=16;
@@ -287,23 +287,23 @@
     while (s<mm_end)
     {
         __asm__ volatile(
-        PREFETCH"  32%1         \n\t"
-        "movq        %1, %%mm0  \n\t"
-        "movq       8%1, %%mm2  \n\t"
-        "movq     %%mm0, %%mm1  \n\t"
-        "movq     %%mm2, %%mm3  \n\t"
-        "psrlq       $1, %%mm0  \n\t"
-        "psrlq       $1, %%mm2  \n\t"
-        "pand     %%mm7, %%mm0  \n\t"
-        "pand     %%mm7, %%mm2  \n\t"
-        "pand     %%mm6, %%mm1  \n\t"
-        "pand     %%mm6, %%mm3  \n\t"
-        "por      %%mm1, %%mm0  \n\t"
-        "por      %%mm3, %%mm2  \n\t"
-        MOVNTQ"   %%mm0,  %0    \n\t"
-        MOVNTQ"   %%mm2, 8%0"
-        :"=m"(*d)
-        :"m"(*s)
+            PREFETCH"  32%1         \n\t"
+            "movq        %1, %%mm0  \n\t"
+            "movq       8%1, %%mm2  \n\t"
+            "movq     %%mm0, %%mm1  \n\t"
+            "movq     %%mm2, %%mm3  \n\t"
+            "psrlq       $1, %%mm0  \n\t"
+            "psrlq       $1, %%mm2  \n\t"
+            "pand     %%mm7, %%mm0  \n\t"
+            "pand     %%mm7, %%mm2  \n\t"
+            "pand     %%mm6, %%mm1  \n\t"
+            "pand     %%mm6, %%mm3  \n\t"
+            "por      %%mm1, %%mm0  \n\t"
+            "por      %%mm3, %%mm2  \n\t"
+            MOVNTQ"   %%mm0,  %0    \n\t"
+            MOVNTQ"   %%mm2, 8%0"
+            :"=m"(*d)
+            :"m"(*s)
         );
         d+=16;
         s+=16;
@@ -339,38 +339,38 @@
     mm_end = end - 15;
 #if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
     __asm__ volatile(
-    "movq           %3, %%mm5   \n\t"
-    "movq           %4, %%mm6   \n\t"
-    "movq           %5, %%mm7   \n\t"
-    "jmp 2f                     \n\t"
-    ASMALIGN(4)
-    "1:                         \n\t"
-    PREFETCH"   32(%1)          \n\t"
-    "movd         (%1), %%mm0   \n\t"
-    "movd        4(%1), %%mm3   \n\t"
-    "punpckldq   8(%1), %%mm0   \n\t"
-    "punpckldq  12(%1), %%mm3   \n\t"
-    "movq        %%mm0, %%mm1   \n\t"
-    "movq        %%mm3, %%mm4   \n\t"
-    "pand        %%mm6, %%mm0   \n\t"
-    "pand        %%mm6, %%mm3   \n\t"
-    "pmaddwd     %%mm7, %%mm0   \n\t"
-    "pmaddwd     %%mm7, %%mm3   \n\t"
-    "pand        %%mm5, %%mm1   \n\t"
-    "pand        %%mm5, %%mm4   \n\t"
-    "por         %%mm1, %%mm0   \n\t"
-    "por         %%mm4, %%mm3   \n\t"
-    "psrld          $5, %%mm0   \n\t"
-    "pslld         $11, %%mm3   \n\t"
-    "por         %%mm3, %%mm0   \n\t"
-    MOVNTQ"      %%mm0, (%0)    \n\t"
-    "add           $16,  %1     \n\t"
-    "add            $8,  %0     \n\t"
-    "2:                         \n\t"
-    "cmp            %2,  %1     \n\t"
-    " jb            1b          \n\t"
-    : "+r" (d), "+r"(s)
-    : "r" (mm_end), "m" (mask3216g), "m" (mask3216br), "m" (mul3216)
+        "movq           %3, %%mm5   \n\t"
+        "movq           %4, %%mm6   \n\t"
+        "movq           %5, %%mm7   \n\t"
+        "jmp 2f                     \n\t"
+        ASMALIGN(4)
+        "1:                         \n\t"
+        PREFETCH"   32(%1)          \n\t"
+        "movd         (%1), %%mm0   \n\t"
+        "movd        4(%1), %%mm3   \n\t"
+        "punpckldq   8(%1), %%mm0   \n\t"
+        "punpckldq  12(%1), %%mm3   \n\t"
+        "movq        %%mm0, %%mm1   \n\t"
+        "movq        %%mm3, %%mm4   \n\t"
+        "pand        %%mm6, %%mm0   \n\t"
+        "pand        %%mm6, %%mm3   \n\t"
+        "pmaddwd     %%mm7, %%mm0   \n\t"
+        "pmaddwd     %%mm7, %%mm3   \n\t"
+        "pand        %%mm5, %%mm1   \n\t"
+        "pand        %%mm5, %%mm4   \n\t"
+        "por         %%mm1, %%mm0   \n\t"
+        "por         %%mm4, %%mm3   \n\t"
+        "psrld          $5, %%mm0   \n\t"
+        "pslld         $11, %%mm3   \n\t"
+        "por         %%mm3, %%mm0   \n\t"
+        MOVNTQ"      %%mm0, (%0)    \n\t"
+        "add           $16,  %1     \n\t"
+        "add            $8,  %0     \n\t"
+        "2:                         \n\t"
+        "cmp            %2,  %1     \n\t"
+        " jb            1b          \n\t"
+        : "+r" (d), "+r"(s)
+        : "r" (mm_end), "m" (mask3216g), "m" (mask3216br), "m" (mul3216)
     );
 #else
     __asm__ volatile(PREFETCH"    %0"::"m"(*src):"memory");
@@ -381,35 +381,35 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movd          %1, %%mm0    \n\t"
-        "movd         4%1, %%mm3    \n\t"
-        "punpckldq    8%1, %%mm0    \n\t"
-        "punpckldq   12%1, %%mm3    \n\t"
-        "movq       %%mm0, %%mm1    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm3, %%mm4    \n\t"
-        "movq       %%mm3, %%mm5    \n\t"
-        "psrlq         $3, %%mm0    \n\t"
-        "psrlq         $3, %%mm3    \n\t"
-        "pand          %2, %%mm0    \n\t"
-        "pand          %2, %%mm3    \n\t"
-        "psrlq         $5, %%mm1    \n\t"
-        "psrlq         $5, %%mm4    \n\t"
-        "pand       %%mm6, %%mm1    \n\t"
-        "pand       %%mm6, %%mm4    \n\t"
-        "psrlq         $8, %%mm2    \n\t"
-        "psrlq         $8, %%mm5    \n\t"
-        "pand       %%mm7, %%mm2    \n\t"
-        "pand       %%mm7, %%mm5    \n\t"
-        "por        %%mm1, %%mm0    \n\t"
-        "por        %%mm4, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm5, %%mm3    \n\t"
-        "psllq        $16, %%mm3    \n\t"
-        "por        %%mm3, %%mm0    \n\t"
-        MOVNTQ"     %%mm0, %0       \n\t"
-        :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            PREFETCH"    32%1           \n\t"
+            "movd          %1, %%mm0    \n\t"
+            "movd         4%1, %%mm3    \n\t"
+            "punpckldq    8%1, %%mm0    \n\t"
+            "punpckldq   12%1, %%mm3    \n\t"
+            "movq       %%mm0, %%mm1    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm3, %%mm4    \n\t"
+            "movq       %%mm3, %%mm5    \n\t"
+            "psrlq         $3, %%mm0    \n\t"
+            "psrlq         $3, %%mm3    \n\t"
+            "pand          %2, %%mm0    \n\t"
+            "pand          %2, %%mm3    \n\t"
+            "psrlq         $5, %%mm1    \n\t"
+            "psrlq         $5, %%mm4    \n\t"
+            "pand       %%mm6, %%mm1    \n\t"
+            "pand       %%mm6, %%mm4    \n\t"
+            "psrlq         $8, %%mm2    \n\t"
+            "psrlq         $8, %%mm5    \n\t"
+            "pand       %%mm7, %%mm2    \n\t"
+            "pand       %%mm7, %%mm5    \n\t"
+            "por        %%mm1, %%mm0    \n\t"
+            "por        %%mm4, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm5, %%mm3    \n\t"
+            "psllq        $16, %%mm3    \n\t"
+            "por        %%mm3, %%mm0    \n\t"
+            MOVNTQ"     %%mm0, %0       \n\t"
+            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
         d += 4;
         s += 16;
     }
@@ -443,35 +443,35 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movd          %1, %%mm0    \n\t"
-        "movd         4%1, %%mm3    \n\t"
-        "punpckldq    8%1, %%mm0    \n\t"
-        "punpckldq   12%1, %%mm3    \n\t"
-        "movq       %%mm0, %%mm1    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm3, %%mm4    \n\t"
-        "movq       %%mm3, %%mm5    \n\t"
-        "psllq         $8, %%mm0    \n\t"
-        "psllq         $8, %%mm3    \n\t"
-        "pand       %%mm7, %%mm0    \n\t"
-        "pand       %%mm7, %%mm3    \n\t"
-        "psrlq         $5, %%mm1    \n\t"
-        "psrlq         $5, %%mm4    \n\t"
-        "pand       %%mm6, %%mm1    \n\t"
-        "pand       %%mm6, %%mm4    \n\t"
-        "psrlq        $19, %%mm2    \n\t"
-        "psrlq        $19, %%mm5    \n\t"
-        "pand          %2, %%mm2    \n\t"
-        "pand          %2, %%mm5    \n\t"
-        "por        %%mm1, %%mm0    \n\t"
-        "por        %%mm4, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm5, %%mm3    \n\t"
-        "psllq        $16, %%mm3    \n\t"
-        "por        %%mm3, %%mm0    \n\t"
-        MOVNTQ"     %%mm0, %0       \n\t"
-        :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            PREFETCH"    32%1           \n\t"
+            "movd          %1, %%mm0    \n\t"
+            "movd         4%1, %%mm3    \n\t"
+            "punpckldq    8%1, %%mm0    \n\t"
+            "punpckldq   12%1, %%mm3    \n\t"
+            "movq       %%mm0, %%mm1    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm3, %%mm4    \n\t"
+            "movq       %%mm3, %%mm5    \n\t"
+            "psllq         $8, %%mm0    \n\t"
+            "psllq         $8, %%mm3    \n\t"
+            "pand       %%mm7, %%mm0    \n\t"
+            "pand       %%mm7, %%mm3    \n\t"
+            "psrlq         $5, %%mm1    \n\t"
+            "psrlq         $5, %%mm4    \n\t"
+            "pand       %%mm6, %%mm1    \n\t"
+            "pand       %%mm6, %%mm4    \n\t"
+            "psrlq        $19, %%mm2    \n\t"
+            "psrlq        $19, %%mm5    \n\t"
+            "pand          %2, %%mm2    \n\t"
+            "pand          %2, %%mm5    \n\t"
+            "por        %%mm1, %%mm0    \n\t"
+            "por        %%mm4, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm5, %%mm3    \n\t"
+            "psllq        $16, %%mm3    \n\t"
+            "por        %%mm3, %%mm0    \n\t"
+            MOVNTQ"     %%mm0, %0       \n\t"
+            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
         d += 4;
         s += 16;
     }
@@ -498,38 +498,38 @@
     mm_end = end - 15;
 #if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
     __asm__ volatile(
-    "movq           %3, %%mm5   \n\t"
-    "movq           %4, %%mm6   \n\t"
-    "movq           %5, %%mm7   \n\t"
-    "jmp            2f          \n\t"
-    ASMALIGN(4)
-    "1:                         \n\t"
-    PREFETCH"   32(%1)          \n\t"
-    "movd         (%1), %%mm0   \n\t"
-    "movd        4(%1), %%mm3   \n\t"
-    "punpckldq   8(%1), %%mm0   \n\t"
-    "punpckldq  12(%1), %%mm3   \n\t"
-    "movq        %%mm0, %%mm1   \n\t"
-    "movq        %%mm3, %%mm4   \n\t"
-    "pand        %%mm6, %%mm0   \n\t"
-    "pand        %%mm6, %%mm3   \n\t"
-    "pmaddwd     %%mm7, %%mm0   \n\t"
-    "pmaddwd     %%mm7, %%mm3   \n\t"
-    "pand        %%mm5, %%mm1   \n\t"
-    "pand        %%mm5, %%mm4   \n\t"
-    "por         %%mm1, %%mm0   \n\t"
-    "por         %%mm4, %%mm3   \n\t"
-    "psrld          $6, %%mm0   \n\t"
-    "pslld         $10, %%mm3   \n\t"
-    "por         %%mm3, %%mm0   \n\t"
-    MOVNTQ"      %%mm0, (%0)    \n\t"
-    "add           $16,  %1     \n\t"
-    "add            $8,  %0     \n\t"
-    "2:                         \n\t"
-    "cmp            %2,  %1     \n\t"
-    " jb            1b          \n\t"
-    : "+r" (d), "+r"(s)
-    : "r" (mm_end), "m" (mask3215g), "m" (mask3216br), "m" (mul3215)
+        "movq           %3, %%mm5   \n\t"
+        "movq           %4, %%mm6   \n\t"
+        "movq           %5, %%mm7   \n\t"
+        "jmp            2f          \n\t"
+        ASMALIGN(4)
+        "1:                         \n\t"
+        PREFETCH"   32(%1)          \n\t"
+        "movd         (%1), %%mm0   \n\t"
+        "movd        4(%1), %%mm3   \n\t"
+        "punpckldq   8(%1), %%mm0   \n\t"
+        "punpckldq  12(%1), %%mm3   \n\t"
+        "movq        %%mm0, %%mm1   \n\t"
+        "movq        %%mm3, %%mm4   \n\t"
+        "pand        %%mm6, %%mm0   \n\t"
+        "pand        %%mm6, %%mm3   \n\t"
+        "pmaddwd     %%mm7, %%mm0   \n\t"
+        "pmaddwd     %%mm7, %%mm3   \n\t"
+        "pand        %%mm5, %%mm1   \n\t"
+        "pand        %%mm5, %%mm4   \n\t"
+        "por         %%mm1, %%mm0   \n\t"
+        "por         %%mm4, %%mm3   \n\t"
+        "psrld          $6, %%mm0   \n\t"
+        "pslld         $10, %%mm3   \n\t"
+        "por         %%mm3, %%mm0   \n\t"
+        MOVNTQ"      %%mm0, (%0)    \n\t"
+        "add           $16,  %1     \n\t"
+        "add            $8,  %0     \n\t"
+        "2:                         \n\t"
+        "cmp            %2,  %1     \n\t"
+        " jb            1b          \n\t"
+        : "+r" (d), "+r"(s)
+        : "r" (mm_end), "m" (mask3215g), "m" (mask3216br), "m" (mul3215)
     );
 #else
     __asm__ volatile(PREFETCH"    %0"::"m"(*src):"memory");
@@ -540,35 +540,35 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movd          %1, %%mm0    \n\t"
-        "movd         4%1, %%mm3    \n\t"
-        "punpckldq    8%1, %%mm0    \n\t"
-        "punpckldq   12%1, %%mm3    \n\t"
-        "movq       %%mm0, %%mm1    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm3, %%mm4    \n\t"
-        "movq       %%mm3, %%mm5    \n\t"
-        "psrlq         $3, %%mm0    \n\t"
-        "psrlq         $3, %%mm3    \n\t"
-        "pand          %2, %%mm0    \n\t"
-        "pand          %2, %%mm3    \n\t"
-        "psrlq         $6, %%mm1    \n\t"
-        "psrlq         $6, %%mm4    \n\t"
-        "pand       %%mm6, %%mm1    \n\t"
-        "pand       %%mm6, %%mm4    \n\t"
-        "psrlq         $9, %%mm2    \n\t"
-        "psrlq         $9, %%mm5    \n\t"
-        "pand       %%mm7, %%mm2    \n\t"
-        "pand       %%mm7, %%mm5    \n\t"
-        "por        %%mm1, %%mm0    \n\t"
-        "por        %%mm4, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm5, %%mm3    \n\t"
-        "psllq        $16, %%mm3    \n\t"
-        "por        %%mm3, %%mm0    \n\t"
-        MOVNTQ"     %%mm0, %0       \n\t"
-        :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+            PREFETCH"    32%1           \n\t"
+            "movd          %1, %%mm0    \n\t"
+            "movd         4%1, %%mm3    \n\t"
+            "punpckldq    8%1, %%mm0    \n\t"
+            "punpckldq   12%1, %%mm3    \n\t"
+            "movq       %%mm0, %%mm1    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm3, %%mm4    \n\t"
+            "movq       %%mm3, %%mm5    \n\t"
+            "psrlq         $3, %%mm0    \n\t"
+            "psrlq         $3, %%mm3    \n\t"
+            "pand          %2, %%mm0    \n\t"
+            "pand          %2, %%mm3    \n\t"
+            "psrlq         $6, %%mm1    \n\t"
+            "psrlq         $6, %%mm4    \n\t"
+            "pand       %%mm6, %%mm1    \n\t"
+            "pand       %%mm6, %%mm4    \n\t"
+            "psrlq         $9, %%mm2    \n\t"
+            "psrlq         $9, %%mm5    \n\t"
+            "pand       %%mm7, %%mm2    \n\t"
+            "pand       %%mm7, %%mm5    \n\t"
+            "por        %%mm1, %%mm0    \n\t"
+            "por        %%mm4, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm5, %%mm3    \n\t"
+            "psllq        $16, %%mm3    \n\t"
+            "por        %%mm3, %%mm0    \n\t"
+            MOVNTQ"     %%mm0, %0       \n\t"
+            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
         d += 4;
         s += 16;
     }
@@ -602,35 +602,35 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movd          %1, %%mm0    \n\t"
-        "movd         4%1, %%mm3    \n\t"
-        "punpckldq    8%1, %%mm0    \n\t"
-        "punpckldq   12%1, %%mm3    \n\t"
-        "movq       %%mm0, %%mm1    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm3, %%mm4    \n\t"
-        "movq       %%mm3, %%mm5    \n\t"
-        "psllq         $7, %%mm0    \n\t"
-        "psllq         $7, %%mm3    \n\t"
-        "pand       %%mm7, %%mm0    \n\t"
-        "pand       %%mm7, %%mm3    \n\t"
-        "psrlq         $6, %%mm1    \n\t"
-        "psrlq         $6, %%mm4    \n\t"
-        "pand       %%mm6, %%mm1    \n\t"
-        "pand       %%mm6, %%mm4    \n\t"
-        "psrlq        $19, %%mm2    \n\t"
-        "psrlq        $19, %%mm5    \n\t"
-        "pand          %2, %%mm2    \n\t"
-        "pand          %2, %%mm5    \n\t"
-        "por        %%mm1, %%mm0    \n\t"
-        "por        %%mm4, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm5, %%mm3    \n\t"
-        "psllq        $16, %%mm3    \n\t"
-        "por        %%mm3, %%mm0    \n\t"
-        MOVNTQ"     %%mm0, %0       \n\t"
-        :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+            PREFETCH"    32%1           \n\t"
+            "movd          %1, %%mm0    \n\t"
+            "movd         4%1, %%mm3    \n\t"
+            "punpckldq    8%1, %%mm0    \n\t"
+            "punpckldq   12%1, %%mm3    \n\t"
+            "movq       %%mm0, %%mm1    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm3, %%mm4    \n\t"
+            "movq       %%mm3, %%mm5    \n\t"
+            "psllq         $7, %%mm0    \n\t"
+            "psllq         $7, %%mm3    \n\t"
+            "pand       %%mm7, %%mm0    \n\t"
+            "pand       %%mm7, %%mm3    \n\t"
+            "psrlq         $6, %%mm1    \n\t"
+            "psrlq         $6, %%mm4    \n\t"
+            "pand       %%mm6, %%mm1    \n\t"
+            "pand       %%mm6, %%mm4    \n\t"
+            "psrlq        $19, %%mm2    \n\t"
+            "psrlq        $19, %%mm5    \n\t"
+            "pand          %2, %%mm2    \n\t"
+            "pand          %2, %%mm5    \n\t"
+            "por        %%mm1, %%mm0    \n\t"
+            "por        %%mm4, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm5, %%mm3    \n\t"
+            "psllq        $16, %%mm3    \n\t"
+            "por        %%mm3, %%mm0    \n\t"
+            MOVNTQ"     %%mm0, %0       \n\t"
+            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
         d += 4;
         s += 16;
     }
@@ -663,35 +663,35 @@
     while (s < mm_end)
     {
         __asm__ volatile(
-        PREFETCH"    32%1           \n\t"
-        "movd          %1, %%mm0    \n\t"
-        "movd         3%1, %%mm3    \n\t"
-        "punpckldq    6%1, %%mm0    \n\t"
-        "punpckldq    9%1, %%mm3    \n\t"
-        "movq       %%mm0, %%mm1    \n\t"
-        "movq       %%mm0, %%mm2    \n\t"
-        "movq       %%mm3, %%mm4    \n\t"
-        "movq       %%mm3, %%mm5    \n\t"
-        "psrlq         $3, %%mm0    \n\t"
-        "psrlq         $3, %%mm3    \n\t"
-        "pand          %2, %%mm0    \n\t"
-        "pand          %2, %%mm3    \n\t"
-        "psrlq         $5, %%mm1    \n\t"
-        "psrlq         $5, %%mm4    \n\t"
-        "pand       %%mm6, %%mm1    \n\t"
-        "pand       %%mm6, %%mm4    \n\t"
-        "psrlq         $8, %%mm2    \n\t"
-        "psrlq         $8, %%mm5    \n\t"
-        "pand       %%mm7, %%mm2    \n\t"
-        "pand       %%mm7, %%mm5    \n\t"
-        "por        %%mm1, %%mm0    \n\t"
-        "por        %%mm4, %%mm3    \n\t"
-        "por        %%mm2, %%mm0    \n\t"
-        "por        %%mm5, %%mm3    \n\t"
-        "psllq        $16, %%mm3    \n\t"
-        "por        %%mm3, %%mm0    \n\t"
-        MOVNTQ"     %%mm0, %0       \n\t"
-        :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            PREFETCH"    32%1           \n\t"
+            "movd          %1, %%mm0    \n\t"
+            "movd         3%1, %%mm3    \n\t"
+            "punpckldq    6%1, %%mm0    \n\t"
+            "punpckldq    9%1, %%mm3    \n\t"
+            "movq       %%mm0, %%mm1    \n\t"
+            "movq       %%mm0, %%mm2    \n\t"
+            "movq       %%mm3, %%mm4    \n\t"
+            "movq       %%mm3, %%mm5    \n\t"
+            "psrlq         $3, %%mm0    \n\t"
+            "psrlq         $3, %%mm3    \n\t"
+            "pand          %2, %%mm0    \n\t"
+            "pand          %2, %%mm3    \n\t"
+            "psrlq         $5, %%mm1    \n\t"
+            "psrlq         $5, %%mm4    \n\t"
+            "pand       %%mm6, %%mm1    \n\t"
+            "pand       %%mm6, %%mm4    \n\t"
+            "psrlq         $8, %%mm2    \n\t"
+            "psrlq         $8, %%mm5    \n\t"
+            "pand       %%mm7, %%mm2    \n\t"
+            "pand       %%mm7, %%mm5    \n\t"
+            "por        %%mm1, %%mm0    \n\t"
+            "por        %%mm4, %%mm3    \n\t"
+            "por        %%mm2, %%mm0    \n\t"
+            "por        %%mm5, %%mm3    \n\t"
+            "psllq        $16, %%mm3    \n\t"
+            "por        %%mm3, %%mm0    \n\t"
+            MOVNTQ"     %%mm0, %0       \n\t"
+            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
         d += 4;
         s += 12;
     }
@@ -726,35 +726,35 @@
     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");
+            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;
     }
@@ -789,35 +789,35 @@
     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");
+            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;
     }
@@ -852,35 +852,35 @@
     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");
+            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;
     }
@@ -932,117 +932,117 @@
     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");
+            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");
+            "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;
     }
@@ -1074,116 +1074,116 @@
     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");
+            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");
+            "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;
     }
@@ -1236,20 +1236,20 @@
     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"
-        PACK_RGB32
-        :"=m"(*d)
-        :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r)
-        :"memory");
+            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"
+            PACK_RGB32
+            :"=m"(*d)
+            :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r)
+            :"memory");
         d += 16;
         s += 4;
     }
@@ -1291,20 +1291,20 @@
     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"
-        PACK_RGB32
-        :"=m"(*d)
-        :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r)
-        :"memory");
+            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"
+            PACK_RGB32
+            :"=m"(*d)
+            :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r)
+            :"memory");
         d += 16;
         s += 4;
     }
@@ -1336,55 +1336,55 @@
     uint8_t *d = dst-idx;
 #if 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"
+        "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"
 # if 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");
+        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 = *(const uint32_t *)&s[idx], g = v & 0xff00ff00;
@@ -1399,47 +1399,47 @@
 #if HAVE_MMX
     x86_reg 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)
+        "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");
@@ -1474,37 +1474,37 @@
 #if HAVE_MMX
 //FIXME handle 2 lines at once (fewer prefetches, reuse some chroma, but very likely memory-limited 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
+            "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
 
@@ -1597,9 +1597,9 @@
         dst  += dstStride;
     }
 #if HAVE_MMX
-__asm__(    EMMS"       \n\t"
-        SFENCE"     \n\t"
-        :::"memory");
+    __asm__(EMMS"       \n\t"
+            SFENCE"     \n\t"
+            :::"memory");
 #endif
 }
 
@@ -1626,37 +1626,37 @@
 #if HAVE_MMX
 //FIXME handle 2 lines at once (fewer prefetches, reuse some chroma, but very likely memory-limited 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
+            "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
@@ -1703,9 +1703,9 @@
         dst += dstStride;
     }
 #if HAVE_MMX
-__asm__(    EMMS"       \n\t"
-        SFENCE"     \n\t"
-        :::"memory");
+    __asm__(EMMS"       \n\t"
+            SFENCE"     \n\t"
+            :::"memory");
 #endif
 }
 
@@ -1755,85 +1755,85 @@
     {
 #if 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
+            "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
+            "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;
@@ -1859,9 +1859,9 @@
         src  += srcStride;
     }
 #if HAVE_MMX
-__asm__ volatile(   EMMS"       \n\t"
-                SFENCE"     \n\t"
-                :::"memory");
+    __asm__ volatile(EMMS"       \n\t"
+                     SFENCE"     \n\t"
+                     :::"memory");
 #endif
 }
 
@@ -1888,51 +1888,51 @@
     }
     dst[2*srcWidth-1]= src[srcWidth-1];
 
-        dst+= dstStride;
+    dst+= dstStride;
 
     for (y=1; y<srcHeight; y++){
 #if HAVE_MMX2 || HAVE_AMD3DNOW
         const x86_reg 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"
+            "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
@@ -1971,9 +1971,9 @@
 #endif
 
 #if HAVE_MMX
-__asm__ volatile(   EMMS"       \n\t"
-                SFENCE"     \n\t"
-                :::"memory");
+    __asm__ volatile(EMMS"       \n\t"
+                     SFENCE"     \n\t"
+                     :::"memory");
 #endif
 }
 
@@ -1993,85 +1993,85 @@
     {
 #if 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" // UYVY UYVY(0)
-        "movq      8(%0, %%"REG_a", 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, %%"REG_a", 2) \n\t"
-
-        "movq     16(%0, %%"REG_a", 4), %%mm1   \n\t" // UYVY UYVY(8)
-        "movq     24(%0, %%"REG_a", 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, %%"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
+            "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" // UYVY UYVY(0)
+            "movq      8(%0, %%"REG_a", 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, %%"REG_a", 2) \n\t"
+
+            "movq     16(%0, %%"REG_a", 4), %%mm1   \n\t" // UYVY UYVY(8)
+            "movq     24(%0, %%"REG_a", 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, %%"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)
-        "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, %%"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
+            "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)
+            "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, %%"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;
@@ -2097,9 +2097,9 @@
         src  += srcStride;
     }
 #if HAVE_MMX
-__asm__ volatile(   EMMS"       \n\t"
-                SFENCE"     \n\t"
-                :::"memory");
+    __asm__ volatile(EMMS"       \n\t"
+                     SFENCE"     \n\t"
+                     :::"memory");
 #endif
 }
 
@@ -2123,225 +2123,225 @@
         for (i=0; i<2; i++)
         {
             __asm__ volatile(
-            "mov                        %2, %%"REG_a"   \n\t"
-            "movq  "MANGLE(ff_bgr2YCoeff)", %%mm6       \n\t"
-            "movq       "MANGLE(ff_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"
+                "mov                        %2, %%"REG_a"   \n\t"
+                "movq  "MANGLE(ff_bgr2YCoeff)", %%mm6       \n\t"
+                "movq       "MANGLE(ff_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(ff_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" ((x86_reg)-width)
-            : "%"REG_a, "%"REG_d
+                "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(ff_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" ((x86_reg)-width)
+                : "%"REG_a, "%"REG_d
             );
             ydst += lumStride;
             src  += srcStride;
         }
         src -= srcStride*2;
         __asm__ volatile(
-        "mov                        %4, %%"REG_a"   \n\t"
-        "movq       "MANGLE(ff_w1111)", %%mm5       \n\t"
-        "movq  "MANGLE(ff_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"
+            "mov                        %4, %%"REG_a"   \n\t"
+            "movq       "MANGLE(ff_w1111)", %%mm5       \n\t"
+            "movq  "MANGLE(ff_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 HAVE_MMX2 || HAVE_AMD3DNOW
-        "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(ff_bgr2VCoeff)", %%mm1       \n\t"
-        "movq  "MANGLE(ff_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(ff_bgr2VCoeff)", %%mm1       \n\t"
+            "movq  "MANGLE(ff_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 HAVE_MMX2 || HAVE_AMD3DNOW
-        "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(ff_w1111)", %%mm5       \n\t"
-        "psrlw                      $2, %%mm4       \n\t"
-        "psrlw                      $2, %%mm2       \n\t"
-#endif
-        "movq  "MANGLE(ff_bgr2VCoeff)", %%mm1       \n\t"
-        "movq  "MANGLE(ff_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(ff_w1111)", %%mm5       \n\t"
+            "psrlw                      $2, %%mm4       \n\t"
+            "psrlw                      $2, %%mm2       \n\t"
+#endif
+            "movq  "MANGLE(ff_bgr2VCoeff)", %%mm1       \n\t"
+            "movq  "MANGLE(ff_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(ff_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
+            "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(ff_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;
@@ -2349,9 +2349,9 @@
         src  += srcStride*2;
     }
 
-    __asm__ volatile(   EMMS"       \n\t"
-                    SFENCE"     \n\t"
-                    :::"memory");
+    __asm__ volatile(EMMS"       \n\t"
+                     SFENCE"     \n\t"
+                     :::"memory");
 #else
     y=0;
 #endif
@@ -2418,48 +2418,48 @@
 #if HAVE_MMX
 #if 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" ((x86_reg)width-15)
-        : "memory", "%"REG_a""
+            "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" ((x86_reg)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" ((x86_reg)width-15)
-        : "memory", "%"REG_a
+            "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" ((x86_reg)width-15)
+            : "memory", "%"REG_a
         );
 #endif
         for (w= (width&(~15)); w < width; w++)
@@ -2480,10 +2480,10 @@
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
-        );
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
+            );
 #endif
 }
 
@@ -2498,95 +2498,95 @@
     w=width/2; h=height/2;
 #if HAVE_MMX
     __asm__ volatile(
-    PREFETCH" %0    \n\t"
-    PREFETCH" %1    \n\t"
-    ::"m"(*(src1+srcStride1)),"m"(*(src2+srcStride2)):"memory");
+        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;
-#if 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];
+        const uint8_t* s1=src1+srcStride1*(y>>1);
+        uint8_t* d=dst1+dstStride1*y;
+        x=0;
+#if 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 (y=0;y<h;y++){
-    const uint8_t* s2=src2+srcStride2*(y>>1);
-    uint8_t* d=dst2+dstStride2*y;
-    x=0;
-#if 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];
+        const uint8_t* s2=src2+srcStride2*(y>>1);
+        uint8_t* d=dst2+dstStride2*y;
+        x=0;
+#if 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];
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }
@@ -2601,84 +2601,84 @@
     long y,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;
-#if 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];
-    }
+        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;
+#if 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];
+        }
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }
@@ -2947,9 +2947,9 @@
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }
@@ -2972,9 +2972,9 @@
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }
@@ -2999,9 +2999,9 @@
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }
@@ -3024,9 +3024,9 @@
     }
 #if HAVE_MMX
     __asm__(
-        EMMS"       \n\t"
-        SFENCE"     \n\t"
-        ::: "memory"
+            EMMS"       \n\t"
+            SFENCE"     \n\t"
+            ::: "memory"
         );
 #endif
 }