Add MSA optimized ARGB4444ToI420 and ARGB4444ToARGB functions

include/libyuv/macros_msa.h

 /*
  *  Copyright 2016 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.
  */
 
 #ifndef INCLUDE_LIBYUV_MACROS_MSA_H_
 #define INCLUDE_LIBYUV_MACROS_MSA_H_
 
 #if !defined(LIBYUV_DISABLE_MSA) && defined(__mips_msa)
 #include <stdint.h>
 #include <msa.h>
 
 #define LD_B(RTYPE, psrc) *((RTYPE*)(psrc))   /* NOLINT */
 #define LD_UB(...) LD_B(v16u8, __VA_ARGS__)
 
+#define LD_H(RTYPE, psrc) *((RTYPE *) (psrc))
+
 #define ST_B(RTYPE, in, pdst) *((RTYPE*)(pdst)) = (in)  /* NOLINT */
 #define ST_UB(...) ST_B(v16u8, __VA_ARGS__)
 
 /* Description : Load two vectors with 16 'byte' sized elements
    Arguments   : Inputs  - psrc, stride
                  Outputs - out0, out1
                  Return Type - as per RTYPE
    Details     : Load 16 byte elements in 'out0' from (psrc)
                  Load 16 byte elements in 'out1' from (psrc + stride)
 */
 #define LD_B2(RTYPE, psrc, stride, out0, out1) {  \
   out0 = LD_B(RTYPE, (psrc));                     \
   out1 = LD_B(RTYPE, (psrc) + stride);            \
 }
 #define LD_UB2(...) LD_B2(v16u8, __VA_ARGS__)
 
 #define LD_B4(RTYPE, psrc, stride, out0, out1, out2, out3) {  \
   LD_B2(RTYPE, (psrc), stride, out0, out1);                   \
   LD_B2(RTYPE, (psrc) + 2 * stride , stride, out2, out3);     \
 }
 #define LD_UB4(...) LD_B4(v16u8, __VA_ARGS__)
 
+/* Description : Load vectors with 8 halfword elements with stride
+   Arguments   : Inputs  - psrc, stride
+                 Outputs - out0, out1
+   Details     : Load 8 halfword elements in 'out0' from (psrc)
+                 Load 8 halfword elements in 'out1' from (psrc + stride)
+*/
+#define LD_H2(RTYPE, psrc, stride, out0, out1) {  \

[fbarchard1, 2016/10/14 21:35:16]

would it more clear in the calling code to do the 2 loads?

[manojkumar.bhosale, 2016/10/19 11:56:27]

Removed this macro

+  out0 = LD_H(RTYPE, (psrc));                     \
+  out1 = LD_H(RTYPE, (psrc) + (stride));          \
+}
+#define LD_UH2(...) LD_H2(v8u16, __VA_ARGS__)
+
 /* Description : Store two vectors with stride each having 16 'byte' sized
                  elements
    Arguments   : Inputs - in0, in1, pdst, stride
    Details     : Store 16 byte elements from 'in0' to (pdst)
                  Store 16 byte elements from 'in1' to (pdst + stride)
 */
 #define ST_B2(RTYPE, in0, in1, pdst, stride) {  \
   ST_B(RTYPE, in0, (pdst));                     \
   ST_B(RTYPE, in1, (pdst) + stride);            \
 }
@@ skipping 27 matching lines @@
 */
 #define ILVRL_B2(RTYPE, in0, in1, out0, out1) {           \
   out0 = (RTYPE) __msa_ilvr_b((v16i8) in0, (v16i8) in1);  \
   out1 = (RTYPE) __msa_ilvl_b((v16i8) in0, (v16i8) in1);  \
 }
 #define ILVRL_B2_UB(...) ILVRL_B2(v16u8, __VA_ARGS__)
 
 #endif  /* !defined(LIBYUV_DISABLE_MSA) && defined(__mips_msa) */
 
 #endif  // INCLUDE_LIBYUV_MACROS_MSA_H_