port ARGB to 565 dithering SSE2 code to GCC.

source/row_gcc.cc

 // VERSION 2
 /*
  *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
@@ skipping 508 matching lines @@
     "lea       " MEMLEA(0x8,1) ",%1            \n"
     "sub       $0x4,%2                         \n"
     "jg        1b                              \n"
   : "+r"(src),  // %0
     "+r"(dst),  // %1
     "+r"(pix)   // %2
   :: "memory", "cc", "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 
+void ARGBToRGB565DitherRow_SSE2(const uint8* src, uint8* dst,
+                                const uint32 dither4, int pix) {
+  asm volatile (
+    "movd       %3,%%xmm6                      \n"
+    "punpcklbw  %%xmm6,%%xmm6                  \n"
+    "movdqa     %%xmm6,%%xmm7                  \n"
+    "punpcklwd  %%xmm6,%%xmm6                  \n"
+    "punpckhwd  %%xmm7,%%xmm7                  \n"
+    "pcmpeqb    %%xmm3,%%xmm3                  \n"
+    "psrld      $0x1b,%%xmm3                   \n"
+    "pcmpeqb    %%xmm4,%%xmm4                  \n"
+    "psrld      $0x1a,%%xmm4                   \n"
+    "pslld      $0x5,%%xmm4                    \n"
+    "pcmpeqb    %%xmm5,%%xmm5                  \n"
+    "pslld      $0xb,%%xmm5                    \n"
+
+    LABELALIGN
+  "1:                                          \n"
+    "movdqu     (%0),%%xmm0                    \n"
+    "paddusb    %%xmm6,%%xmm0                  \n"
+    "movdqa     %%xmm0,%%xmm1                  \n"
+    "movdqa     %%xmm0,%%xmm2                  \n"
+    "pslld      $0x8,%%xmm0                    \n"
+    "psrld      $0x3,%%xmm1                    \n"
+    "psrld      $0x5,%%xmm2                    \n"
+    "psrad      $0x10,%%xmm0                   \n"
+    "pand       %%xmm3,%%xmm1                  \n"
+    "pand       %%xmm4,%%xmm2                  \n"
+    "pand       %%xmm5,%%xmm0                  \n"
+    "por        %%xmm2,%%xmm1                  \n"
+    "por        %%xmm1,%%xmm0                  \n"
+    "packssdw   %%xmm0,%%xmm0                  \n"
+    "lea        0x10(%0),%0                    \n"
+    "movq       %%xmm0,(%1)                    \n"
+    "lea        0x8(%1),%1                     \n"
+    "sub        $0x4,%2                        \n"
+    "jg        1b                              \n"
+  : "+r"(src),  // %0
+    "+r"(dst),  // %1
+    "+r"(pix)   // %2
+  : "m"(dither4) // %3
+  : "memory", "cc",
+    "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
+  );
+}
+
 void ARGBToARGB1555Row_SSE2(const uint8* src, uint8* dst, int pix) {
   asm volatile (
     "pcmpeqb   %%xmm4,%%xmm4                   \n"
     "psrld     $0x1b,%%xmm4                    \n"
     "movdqa    %%xmm4,%%xmm5                   \n"
     "pslld     $0x5,%%xmm5                     \n"
     "movdqa    %%xmm4,%%xmm6                   \n"
     "pslld     $0xa,%%xmm6                     \n"
     "pcmpeqb   %%xmm7,%%xmm7                   \n"
     "pslld     $0xf,%%xmm7                     \n"
@@ skipping 4953 matching lines @@
   );
 }
 #endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
 
 #endif  // defined(__x86_64__) || defined(__i386__)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif