remove sse2 functions that also have ssse3

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 103 matching lines @@
 #define HAS_I422TORGB565ROW_SSSE3
 #define HAS_I422TORGBAROW_SSSE3
 #define HAS_I422TOUYVYROW_SSE2
 #define HAS_I422TOYUY2ROW_SSE2
 #define HAS_J400TOARGBROW_SSE2
 #define HAS_J422TOARGBROW_SSSE3
 #define HAS_J422TOABGRROW_SSSE3
 #define HAS_H422TOARGBROW_SSSE3
 #define HAS_H422TOABGRROW_SSSE3
 #define HAS_MERGEUVROW_SSE2
-#define HAS_MIRRORROW_SSE2

[harryjin, 2015/09/30 02:42:07]

So for the very few SSE2 only case, things just got a little slower but nothing
breaks right?

 #define HAS_MIRRORROW_SSSE3
 #define HAS_MIRRORROW_UV_SSSE3
 #define HAS_MIRRORUVROW_SSSE3
 #define HAS_NV12TOARGBROW_SSSE3
 #define HAS_NV12TORGB565ROW_SSSE3
 #define HAS_RAWTOARGBROW_SSSE3
 #define HAS_RAWTOYROW_SSSE3
 #define HAS_RGB24TOARGBROW_SSSE3
 #define HAS_RGB24TOYROW_SSSE3
 #define HAS_RGB565TOARGBROW_SSE2
@@ skipping 39 matching lines @@
 #define HAS_INTERPOLATEROW_SSE2
 #define HAS_INTERPOLATEROW_SSSE3
 #define HAS_RGBCOLORTABLEROW_X86
 #define HAS_SOBELROW_SSE2
 #define HAS_SOBELTOPLANEROW_SSE2
 #define HAS_SOBELXROW_SSE2
 #define HAS_SOBELXYROW_SSE2
 #define HAS_SOBELYROW_SSE2
 #endif
 
-// The following are available on x64 Visual C and clangcl.
-// TODO(fbarchard): Port to gcc.
+// The following are also available on x64 Visual C.
 #if !defined(LIBYUV_DISABLE_X86) && defined (_M_X64) && \
     (!defined(__clang__) || defined(__SSSE3__))
 #define HAS_I422TOARGBROW_SSSE3
 #define HAS_I422TOABGRROW_SSSE3
 #define HAS_I422ALPHATOARGBROW_SSSE3
 #define HAS_I422ALPHATOABGRROW_SSSE3
 #endif
 
 // The following are available for AVX2 Visual C and clangcl 32 bit:
 // TODO(fbarchard): Port to gcc.
@@ skipping 59 matching lines @@
 #define HAS_I422ALPHATOABGRROW_AVX2
 
 // Effects:
 #define HAS_ARGBADDROW_AVX2
 #define HAS_ARGBATTENUATEROW_AVX2
 #define HAS_ARGBMULTIPLYROW_AVX2
 #define HAS_ARGBSUBTRACTROW_AVX2
 #define HAS_ARGBUNATTENUATEROW_AVX2
 #endif
 
-// The following are disabled when SSSE3 is available:
-// TODO(fbarchard): remove sse2.  ssse3 is faster and well supported.
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)) && \
-    !defined(LIBYUV_SSSE3_ONLY)
-#define HAS_ARGBATTENUATEROW_SSE2
-#define HAS_ARGBBLENDROW_SSE2
-#define HAS_MIRRORROW_SSE2
-#endif
-
 // The following are available on Neon platforms:
 #if !defined(LIBYUV_DISABLE_NEON) && \
     (defined(__aarch64__) || defined(__ARM_NEON__) || defined(LIBYUV_NEON))
 #define HAS_ABGRTOUVROW_NEON
 #define HAS_ABGRTOYROW_NEON
 #define HAS_ARGB1555TOARGBROW_NEON
 #define HAS_ARGB1555TOUVROW_NEON
 #define HAS_ARGB1555TOYROW_NEON
 #define HAS_ARGB4444TOARGBROW_NEON
 #define HAS_ARGB4444TOUVROW_NEON
@@ skipping 530 matching lines @@
                       uint8* dst_u, uint8* dst_v, int width);
 void ARGBToUV422Row_C(const uint8* src_argb,
                       uint8* dst_u, uint8* dst_v, int width);
 void ARGBToUV411Row_C(const uint8* src_argb,
                       uint8* dst_u, uint8* dst_v, int width);
 void ARGBToUVJ422Row_C(const uint8* src_argb,
                        uint8* dst_u, uint8* dst_v, int width);
 
 void MirrorRow_AVX2(const uint8* src, uint8* dst, int width);
 void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width);
-void MirrorRow_SSE2(const uint8* src, uint8* dst, int width);
 void MirrorRow_NEON(const uint8* src, uint8* dst, int width);
 void MirrorRow_MIPS_DSPR2(const uint8* src, uint8* dst, int width);
 void MirrorRow_C(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_AVX2(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_SSSE3(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_SSE2(const uint8* src, uint8* dst, int width);
 void MirrorRow_Any_NEON(const uint8* src, uint8* dst, int width);
 
 void MirrorUVRow_SSSE3(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
                        int width);
@@ skipping 777 matching lines @@
 void I400ToARGBRow_SSE2(const uint8* src_y, uint8* dst_argb, int width);
 void I400ToARGBRow_AVX2(const uint8* src_y, uint8* dst_argb, int width);
 void I400ToARGBRow_NEON(const uint8* src_y, uint8* dst_argb, int width);
 void I400ToARGBRow_Any_SSE2(const uint8* src_y, uint8* dst_argb, int width);
 void I400ToARGBRow_Any_AVX2(const uint8* src_y, uint8* dst_argb, int width);
 void I400ToARGBRow_Any_NEON(const uint8* src_y, uint8* dst_argb, int width);
 
 // ARGB preattenuated alpha blend.
 void ARGBBlendRow_SSSE3(const uint8* src_argb, const uint8* src_argb1,
                         uint8* dst_argb, int width);
-void ARGBBlendRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width);
 void ARGBBlendRow_NEON(const uint8* src_argb, const uint8* src_argb1,
                        uint8* dst_argb, int width);
 void ARGBBlendRow_C(const uint8* src_argb, const uint8* src_argb1,
                     uint8* dst_argb, int width);
 
 // ARGB multiply images. Same API as Blend, but these require
 // pointer and width alignment for SSE2.
 void ARGBMultiplyRow_C(const uint8* src_argb, const uint8* src_argb1,
                        uint8* dst_argb, int width);
 void ARGBMultiplyRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
@@ skipping 299 matching lines @@
                             const uint8* src_u,
                             const uint8* src_v,
                             uint8* dst_yuy2, int width);
 void I422ToUYVYRow_Any_NEON(const uint8* src_y,
                             const uint8* src_u,
                             const uint8* src_v,
                             uint8* dst_uyvy, int width);
 
 // Effects related row functions.
 void ARGBAttenuateRow_C(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width);
 void ARGBAttenuateRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width);
 void ARGBAttenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb, int width);
 void ARGBAttenuateRow_NEON(const uint8* src_argb, uint8* dst_argb, int width);
 void ARGBAttenuateRow_Any_SSE2(const uint8* src_argb, uint8* dst_argb,
                                int width);
 void ARGBAttenuateRow_Any_SSSE3(const uint8* src_argb, uint8* dst_argb,
                                 int width);
 void ARGBAttenuateRow_Any_AVX2(const uint8* src_argb, uint8* dst_argb,
                                int width);
 void ARGBAttenuateRow_Any_NEON(const uint8* src_argb, uint8* dst_argb,
@@ skipping 160 matching lines @@
 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_  NOLINT