Add MSA optimized ARGB4444ToI420 and ARGB4444ToARGB functions

source/convert.cc

 /*
  *  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 1211 matching lines @@
 }
 
 // Convert ARGB4444 to I420.
 LIBYUV_API
 int ARGB4444ToI420(const uint8* src_argb4444, int src_stride_argb4444,
                    uint8* dst_y, int dst_stride_y,
                    uint8* dst_u, int dst_stride_u,
                    uint8* dst_v, int dst_stride_v,
                    int width, int height) {
   int y;
-#if defined(HAS_ARGB4444TOYROW_NEON)
+#if (defined(HAS_ARGB4444TOYROW_NEON) || defined(HAS_ARGB4444TOYROW_MSA))

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

I'd have suggested taking the intel approach, which write a single function for
this function to convert ARGB4444 to ARGB.  A bit of calling overhead but much
less code.
On Neon the vector load instructions made it easy to write direct loaders for
many formats, but on Intel and Mips a lot of code is needed to read different
RGB formats.

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

Done.

   void (*ARGB4444ToUVRow)(const uint8* src_argb4444, int src_stride_argb4444,
       uint8* dst_u, uint8* dst_v, int width) = ARGB4444ToUVRow_C;
   void (*ARGB4444ToYRow)(const uint8* src_argb4444, uint8* dst_y, int width) =
       ARGB4444ToYRow_C;
 #else
   void (*ARGB4444ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int width) =
       ARGB4444ToARGBRow_C;
   void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
       uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
   void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int width) =
@@ skipping 15 matching lines @@
   if (TestCpuFlag(kCpuHasNEON)) {
     ARGB4444ToUVRow = ARGB4444ToUVRow_Any_NEON;
     ARGB4444ToYRow = ARGB4444ToYRow_Any_NEON;
     if (IS_ALIGNED(width, 8)) {
       ARGB4444ToYRow = ARGB4444ToYRow_NEON;
       if (IS_ALIGNED(width, 16)) {
         ARGB4444ToUVRow = ARGB4444ToUVRow_NEON;
       }
     }
   }
+#elif defined(HAS_ARGB4444TOYROW_MSA)
+  if (TestCpuFlag(kCpuHasMSA)) {
+    ARGB4444ToUVRow = ARGB4444ToUVRow_Any_MSA;
+    ARGB4444ToYRow = ARGB4444ToYRow_Any_MSA;
+    if (IS_ALIGNED(width, 16)) {
+      ARGB4444ToYRow = ARGB4444ToYRow_MSA;
+      if (IS_ALIGNED(width, 32)) {
+        ARGB4444ToUVRow = ARGB4444ToUVRow_MSA;
+      }
+    }
+  }
 // Other platforms do intermediate conversion from ARGB4444 to ARGB.
 #else
 #if defined(HAS_ARGB4444TOARGBROW_SSE2)
   if (TestCpuFlag(kCpuHasSSE2)) {
     ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_SSE2;
     if (IS_ALIGNED(width, 8)) {
       ARGB4444ToARGBRow = ARGB4444ToARGBRow_SSE2;
     }
   }
 #endif
@@ skipping 25 matching lines @@
     }
   }
 #endif
   {
     // Allocate 2 rows of ARGB.
     const int kRowSize = (width * 4 + 31) & ~31;
     align_buffer_64(row, kRowSize * 2);
 #endif
 
     for (y = 0; y < height - 1; y += 2) {
-#if defined(HAS_ARGB4444TOYROW_NEON)
+#if (defined(HAS_ARGB4444TOYROW_NEON) || defined(HAS_ARGB4444TOYROW_MSA))
       ARGB4444ToUVRow(src_argb4444, src_stride_argb4444, dst_u, dst_v, width);
       ARGB4444ToYRow(src_argb4444, dst_y, width);
       ARGB4444ToYRow(src_argb4444 + src_stride_argb4444, dst_y + dst_stride_y,
                      width);
 #else
       ARGB4444ToARGBRow(src_argb4444, row, width);
       ARGB4444ToARGBRow(src_argb4444 + src_stride_argb4444, row + kRowSize,
                         width);
       ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
       ARGBToYRow(row, dst_y, width);
       ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
 #endif
       src_argb4444 += src_stride_argb4444 * 2;
       dst_y += dst_stride_y * 2;
       dst_u += dst_stride_u;
       dst_v += dst_stride_v;
     }
     if (height & 1) {
-#if defined(HAS_ARGB4444TOYROW_NEON)
+#if (defined(HAS_ARGB4444TOYROW_NEON) || defined(HAS_ARGB4444TOYROW_MSA))
       ARGB4444ToUVRow(src_argb4444, 0, dst_u, dst_v, width);
       ARGB4444ToYRow(src_argb4444, dst_y, width);
 #else
       ARGB4444ToARGBRow(src_argb4444, row, width);
       ARGBToUVRow(row, 0, dst_u, dst_v, width);
       ARGBToYRow(row, dst_y, width);
 #endif
     }
-#if !defined(HAS_ARGB4444TOYROW_NEON)
+#if !(defined(HAS_ARGB4444TOYROW_NEON) || defined(HAS_ARGB4444TOYROW_MSA))
     free_aligned_buffer_64(row);
   }
 #endif
   return 0;
 }
 
 static void SplitPixels(const uint8* src_u, int src_pixel_stride_uv,
                         uint8* dst_u, int width) {
   int i;
   for (i = 0; i < width; ++i) {
@@ skipping 65 matching lines @@
     dst_u += dst_stride_u;
     dst_v += dst_stride_v;
   }
   return 0;
 }
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif