Port HalfFloatRow_SSE2 to AVX2 but not using F16C.

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 5323 matching lines @@
     "+r"(dst),    // %1
     "+r"(width)   // %2
   : "x"(scale * kScaleBias)   // %3
   : "memory", "cc",
     "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 #endif  // HAS_HALFFLOATROW_SSE2
 
 #ifdef HAS_HALFFLOATROW_AVX2
+// TODO(fbarchard): consider vadddw instead of vmulps
 void HalfFloatRow_AVX2(const uint16* src, uint16* dst, float scale, int width) {
   asm volatile (
+    "vbroadcastss  %3, %%ymm4                  \n"
+    "vpxor      %%ymm5,%%ymm5,%%ymm5           \n"
+
+    // 16 pixel loop.
+    LABELALIGN
+  "1:                                          \n"
+    "vmovdqu    " MEMACCESS(0) ",%%ymm2        \n"  // 8 shorts
+    "lea        " MEMLEA(0x20,0) ",%0          \n"
+    "vpunpckhwd %%ymm2,%%ymm5,%%ymm3           \n"
+    "vpunpcklwd %%ymm2,%%ymm5,%%ymm2           \n"
+    "vcvtdq2ps  %%ymm3,%%ymm3                  \n"
+    "vcvtdq2ps  %%ymm2,%%ymm2                  \n"
+    "vmulps     %%ymm3,%%ymm4,%%ymm3           \n"
+    "vmulps     %%ymm2,%%ymm4,%%ymm2           \n"
+    "vpsrld     $0xd,%%ymm3,%%ymm3             \n"
+    "vpsrld     $0xd,%%ymm2,%%ymm2             \n"
+    "vpackssdw  %%ymm3, %%ymm2, %%ymm2         \n"  // mutates
+    "vmovdqu    %%ymm2," MEMACCESS(1) "        \n"
+    "lea        " MEMLEA(0x20,1) ",%1          \n"
+    "sub        $0x10,%2                       \n"
+    "jg         1b                             \n"
+    "vzeroupper                                \n"
+  : "+r"(src),    // %0
+    "+r"(dst),    // %1
+    "+r"(width)   // %2
+  : "x"(scale * kScaleBias)   // %3
+  : "memory", "cc",
+    "xmm2", "xmm3", "xmm4", "xmm5"
+  );
+}
+#endif  // HAS_HALFFLOATROW_AVX2
+
+#ifdef HAS_HALFFLOATROW_F16C
+void HalfFloatRow_F16C(const uint16* src, uint16* dst, float scale, int width) {
+  asm volatile (
    "vbroadcastss  %3, %%ymm4                  \n"
 
     // 16 pixel loop.
     LABELALIGN
   "1:                                          \n"
     "vpmovzxwd   " MEMACCESS(0) ",%%ymm2       \n"  // 8 shorts -> 8 ints
     "vpmovzxwd   " MEMACCESS2(0x10,0) ",%%ymm3 \n"  // 8 more
     "lea         " MEMLEA(0x20,0) ",%0         \n"
     "vcvtdq2ps   %%ymm2,%%ymm2                 \n"
     "vcvtdq2ps   %%ymm3,%%ymm3                 \n"
     "vmulps      %%ymm2,%%ymm4,%%ymm2          \n"
     "vmulps      %%ymm3,%%ymm4,%%ymm3          \n"
     "vcvtps2ph   $3, %%ymm2, %%xmm2            \n"
     "vcvtps2ph   $3, %%ymm3, %%xmm3            \n"
     "vmovdqu     %%xmm2," MEMACCESS(1) "       \n"
     "vmovdqu     %%xmm3," MEMACCESS2(0x10,1) " \n"
     "lea         " MEMLEA(0x20,1) ",%1         \n"
     "sub         $0x10,%2                      \n"
     "jg          1b                            \n"
+
     "vzeroupper                                \n"
   : "+r"(src),   // %0
     "+r"(dst),   // %1
     "+r"(width)  // %2
   : "x"(scale)   // %3
   : "memory", "cc",
     "xmm2", "xmm3", "xmm4"
   );
 }
-#endif  // HAS_HALFFLOATROW_AVX2
+#endif  // HAS_HALFFLOATROW_F16C
 
 #ifdef HAS_ARGBCOLORTABLEROW_X86
 // Tranform ARGB pixels with color table.
 void ARGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb,
                            int width) {
   uintptr_t pixel_temp;
   asm volatile (
     // 1 pixel loop.
     LABELALIGN
   "1:                                          \n"
@@ skipping 147 matching lines @@
   );
 }
 #endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
 
 #endif  // defined(__x86_64__) || defined(__i386__)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif