Half float neon port, enable O2 for neon, and denormal test.

include/libyuv/row.h

 /*
  *  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 183 matching lines @@
 #define HAS_ARGBPOLYNOMIALROW_AVX2
 #define HAS_ARGBSHUFFLEROW_AVX2
 #define HAS_ARGBTORGB565DITHERROW_AVX2
 #define HAS_ARGBTOUVJROW_AVX2
 #define HAS_ARGBTOUVROW_AVX2
 #define HAS_ARGBTOYJROW_AVX2
 #define HAS_ARGBTOYROW_AVX2
 #define HAS_COPYROW_AVX
 #define HAS_H422TOARGBROW_AVX2
 #define HAS_HALFFLOATROW_AVX2
-// #define HAS_HALFFLOATROW_F16C  // Enable to test halffloat cast
+//  #define HAS_HALFFLOATROW_F16C  // Enable to test halffloat cast
 #define HAS_I400TOARGBROW_AVX2
 #define HAS_I422TOARGB1555ROW_AVX2
 #define HAS_I422TOARGB4444ROW_AVX2
 #define HAS_I422TOARGBROW_AVX2
 #define HAS_I422TORGB24ROW_AVX2
 #define HAS_I422TORGB565ROW_AVX2
 #define HAS_I422TORGBAROW_AVX2
 #define HAS_I444TOARGBROW_AVX2
 #define HAS_INTERPOLATEROW_AVX2
 #define HAS_J422TOARGBROW_AVX2
@@ skipping 108 matching lines @@
 #define HAS_SPLITUVROW_NEON
 #define HAS_UYVYTOARGBROW_NEON
 #define HAS_UYVYTOUV422ROW_NEON
 #define HAS_UYVYTOUVROW_NEON
 #define HAS_UYVYTOYROW_NEON
 #define HAS_YUY2TOARGBROW_NEON
 #define HAS_YUY2TOUV422ROW_NEON
 #define HAS_YUY2TOUVROW_NEON
 #define HAS_YUY2TOYROW_NEON
 
+// TODO(fbarchard): Port to 32 bit.
+#if defined(__aarch64__)
+#define HAS_HALFFLOATROW_NEON
+#endif
+
 // Effects:
 #define HAS_ARGBADDROW_NEON
 #define HAS_ARGBATTENUATEROW_NEON
 #define HAS_ARGBBLENDROW_NEON
 #define HAS_ARGBCOLORMATRIXROW_NEON
 #define HAS_ARGBGRAYROW_NEON
 #define HAS_ARGBMIRRORROW_NEON
 #define HAS_ARGBMULTIPLYROW_NEON
 #define HAS_ARGBQUANTIZEROW_NEON
 #define HAS_ARGBSEPIAROW_NEON
@@ skipping 1604 matching lines @@
 void HalfFloatRow_C(const uint16* src, uint16* dst, float scale, int width);
 void HalfFloatRow_SSE2(const uint16* src, uint16* dst, float scale, int width);
 void HalfFloatRow_Any_SSE2(const uint16* src, uint16* dst, float scale,
                            int width);
 void HalfFloatRow_AVX2(const uint16* src, uint16* dst, float scale, int width);
 void HalfFloatRow_Any_AVX2(const uint16* src, uint16* dst, float scale,
                            int width);
 void HalfFloatRow_F16C(const uint16* src, uint16* dst, float scale, int width);
 void HalfFloatRow_Any_F16C(const uint16* src, uint16* dst, float scale,
                            int width);
+void HalfFloatRow_NEON(const uint16* src, uint16* dst, float scale, int width);
+void HalfFloatRow_Any_NEON(const uint16* src, uint16* dst, float scale,
+                           int width);
 
 void ARGBLumaColorTableRow_C(const uint8* src_argb, uint8* dst_argb, int width,
                              const uint8* luma, uint32 lumacoeff);
 void ARGBLumaColorTableRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
                                  int width,
                                  const uint8* luma, uint32 lumacoeff);
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif
 
 #endif  // INCLUDE_LIBYUV_ROW_H_