[x265] [PATCH] Adapting weightp primitive for pixel input

[shazeb at multicorewareinc.com]

# HG changeset patch
# User Shazeb Nawaz Khan <shazeb at multicorewareinc.com>
# Date 1382354467 -19800
#      Mon Oct 21 16:51:07 2013 +0530
# Node ID 3e732910246416da437183675b0892f6ebb43005
# Parent  f987c24c7bf2d3e6df0f96b2c75f3e71ab15a85e
Adapting weightp primitive for pixel input


by simulating shift & round as in convertPixelToShort Primitive

diff -r f987c24c7bf2 -r 3e7329102464 source/common/pixel.cpp
--- a/source/common/pixel.cpp	Mon Oct 21 16:15:22 2013 +0530
+++ b/source/common/pixel.cpp	Mon Oct 21 16:51:07 2013 +0530
@@ -516,8 +516,7 @@
     }
 }
 
-template<typename T>
-void weightUnidir(T *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
+void weightUnidir(int16_t *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
 {
     int x, y;
 
@@ -525,15 +524,31 @@
     {
         for (x = width - 1; x >= 0; )
         {
-            // note: luma min width is 4
-            dst[x] = (pixel)Clip3(0, ((1 << X265_DEPTH) - 1), ((w0 * (src[x] + IF_INTERNAL_OFFS) + round) >> shift) + offset);
-            x--;
+            // note: width can be odd
             dst[x] = (pixel)Clip3(0, ((1 << X265_DEPTH) - 1), ((w0 * (src[x] + IF_INTERNAL_OFFS) + round) >> shift) + offset);
             x--;
         }
 
         src += srcStride;
-        dst  += dstStride;
+        dst += dstStride;
+    }
+}
+
+void weightUnidirPix(pixel *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
+{
+    int x, y;
+    for (y = height - 1; y >= 0; y--)
+    {
+        for (x = width - 1; x >= 0; )
+        {
+            // simulating pixel to short conversion
+            short val = src[x] << (IF_INTERNAL_PREC - X265_DEPTH);
+            dst[x] = (pixel) Clip3(0, ((1 << X265_DEPTH) - 1), ((w0 * (val) + round) >> shift) + offset);
+            x--;
+        }
+
+        src += srcStride;
+        dst += dstStride;
     }
 }
 
@@ -856,8 +871,8 @@
     p.transpose[3] = transpose<32>;
     p.transpose[4] = transpose<64>;
 
-    p.weightpUniPixel = weightUnidir<pixel>;
-    p.weightpUni = weightUnidir<int16_t>;
+    p.weightpUniPixel = weightUnidirPix;
+    p.weightpUni = weightUnidir;
 
     p.pixelsub_sp = pixelsub_sp_c;
     p.pixeladd_pp = pixeladd_pp_c;
diff -r f987c24c7bf2 -r 3e7329102464 source/common/vec/pixel-sse41.cpp
--- a/source/common/vec/pixel-sse41.cpp	Mon Oct 21 16:15:22 2013 +0530
+++ b/source/common/vec/pixel-sse41.cpp	Mon Oct 21 16:51:07 2013 +0530
@@ -5659,8 +5659,8 @@
     p.sse_pp[LUMA_64x48] = sse_pp_64<48>;
     p.sse_pp[LUMA_64x16] = sse_pp_64<16>;
 
-    p.weightpUniPixel = weightUnidirPixel;
-    p.weightpUni = weightUnidir;
+//    p.weightpUniPixel = weightUnidirPixel;
+//    p.weightpUni = weightUnidir;
 #endif /* !HIGH_BIT_DEPTH */
 }
 }