[x265] [PATCH] primitive function for luma and chroma for loops in addAvg()

[dnyaneshwar at multicorewareinc.com]

# HG changeset patch
# User Dnyaneshwar Gorade <dnyaneshwar at multicorewareinc.com>
# Date 1384495519 -19800
#      Fri Nov 15 11:35:19 2013 +0530
# Node ID 3c396b064d578b480302e70f52dcc2bf4380f74b
# Parent  c4ca80d19105ccf1ba2ec14dd65915f2820a660d
primitive function for luma and chroma for loops in addAvg()

diff -r c4ca80d19105 -r 3c396b064d57 source/Lib/TLibCommon/TComYuv.cpp
--- a/source/Lib/TLibCommon/TComYuv.cpp	Tue Nov 12 19:10:23 2013 +0530
+++ b/source/Lib/TLibCommon/TComYuv.cpp	Fri Nov 15 11:35:19 2013 +0530
@@ -590,7 +590,6 @@
 
 void TComYuv::addAvg(TShortYUV* srcYuv0, TShortYUV* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma)
 {
-    int x, y;
     uint32_t src0Stride, src1Stride, dststride;
     int shiftNum, offset;
 
@@ -606,6 +605,8 @@
     Pel* dstU = getCbAddr(partUnitIdx);
     Pel* dstV = getCrAddr(partUnitIdx);
 
+    int part = partitionFromSizes(width, height);
+
     if (bLuma)
     {
         src0Stride = srcYuv0->m_width;
@@ -614,20 +615,7 @@
         shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
         offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
 
-        for (y = 0; y < height; y++)
-        {
-            for (x = 0; x < width; x += 4)
-            {
-                dstY[x + 0] = ClipY((srcY0[x + 0] + srcY1[x + 0] + offset) >> shiftNum);
-                dstY[x + 1] = ClipY((srcY0[x + 1] + srcY1[x + 1] + offset) >> shiftNum);
-                dstY[x + 2] = ClipY((srcY0[x + 2] + srcY1[x + 2] + offset) >> shiftNum);
-                dstY[x + 3] = ClipY((srcY0[x + 3] + srcY1[x + 3] + offset) >> shiftNum);
-            }
-
-            srcY0 += src0Stride;
-            srcY1 += src1Stride;
-            dstY  += dststride;
-        }
+        primitives.luma_addAvg[part](dstY, dststride, srcY0, src0Stride, srcY1, src1Stride);
     }
     if (bChroma)
     {
@@ -641,26 +629,8 @@
         width  >>= m_hChromaShift;
         height >>= m_vChromaShift;
 
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstU[x] = ClipC((srcU0[x] + srcU1[x] + offset) >> shiftNum);
-                dstV[x] = ClipC((srcV0[x] + srcV1[x] + offset) >> shiftNum);
-                x--;
-                dstU[x] = ClipC((srcU0[x] + srcU1[x] + offset) >> shiftNum);
-                dstV[x] = ClipC((srcV0[x] + srcV1[x] + offset) >> shiftNum);
-                x--;
-            }
-
-            srcU0 += src0Stride;
-            srcU1 += src1Stride;
-            srcV0 += src0Stride;
-            srcV1 += src1Stride;
-            dstU  += dststride;
-            dstV  += dststride;
-        }
+        primitives.chroma_addAvg[part](dstU, dststride, srcU0, src0Stride, srcU1, src1Stride);
+        primitives.chroma_addAvg[part](dstV, dststride, srcV0, src0Stride, srcV1, src1Stride);
     }
 }
 
diff -r c4ca80d19105 -r 3c396b064d57 source/common/pixel.cpp
--- a/source/common/pixel.cpp	Tue Nov 12 19:10:23 2013 +0530
+++ b/source/common/pixel.cpp	Fri Nov 15 11:35:19 2013 +0530
@@ -794,6 +794,27 @@
         a += dstride;
     }
 }
+
+template<int bx, int by>
+void addAvg(pixel* dst, intptr_t dstStride, int16_t* src0, intptr_t src0Stride, int16_t* src1, intptr_t src1Stride)
+{
+    int shiftNum, offset;
+    shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
+    offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
+
+    for (int y = 0; y < by; y++)
+    {
+        for (int x = 0; x < bx; x += 2)
+        {
+            dst[x + 0] = ClipY((src0[x + 0] + src1[x + 0] + offset) >> shiftNum);
+            dst[x + 1] = ClipY((src0[x + 1] + src1[x + 1] + offset) >> shiftNum);
+        }
+
+        src0 += src0Stride;
+        src1 += src1Stride;
+        dst  += dstStride;
+    }
+}
 }  // end anonymous namespace
 
 namespace x265 {
@@ -835,12 +856,14 @@
     p.satd[LUMA_16x64] = satd8<16, 64>;
 
 #define CHROMA(W, H) \
+    p.chroma_addAvg[CHROMA_ ## W ## x ## H]  = addAvg<W, H>; \
     p.chroma_copy_pp[CHROMA_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
     p.chroma_copy_sp[CHROMA_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
     p.chroma_copy_ps[CHROMA_ ## W ## x ## H] = blockcopy_ps_c<W, H>;\
     p.chroma_sub_ps[CHROMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>;
 
 #define LUMA(W, H) \
+    p.luma_addAvg[LUMA_ ## W ## x ## H]  = addAvg<W, H>; \
     p.luma_copy_pp[LUMA_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
     p.luma_copy_sp[LUMA_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
     p.luma_copy_ps[LUMA_ ## W ## x ## H] = blockcopy_ps_c<W, H>;\
diff -r c4ca80d19105 -r 3c396b064d57 source/common/primitives.h
--- a/source/common/primitives.h	Tue Nov 12 19:10:23 2013 +0530
+++ b/source/common/primitives.h	Fri Nov 15 11:35:19 2013 +0530
@@ -208,6 +208,8 @@
 
 typedef void (*pixel_sub_ps_t)(int16_t *dst, intptr_t dstride, pixel *src0, pixel *src1, intptr_t sstride0, intptr_t sstride1);
 
+typedef void (*addAvg_t)(pixel* dst, intptr_t dstStride, int16_t* src0, intptr_t src0Stride, int16_t* src1, intptr_t src1Stride);
+
 /* Define a structure containing function pointers to optimized encoder
  * primitives.  Each pointer can reference either an assembly routine,
  * a vectorized primitive, or a C function. */
@@ -288,6 +290,9 @@
     var_t           var[NUM_LUMA_PARTITIONS];
     ssim_4x4x2_core_t ssim_4x4x2_core;
     plane_copy_deinterleave_t plane_copy_deinterleave_c;
+
+    addAvg_t        luma_addAvg[NUM_LUMA_PARTITIONS];
+    addAvg_t        chroma_addAvg[NUM_CHROMA_PARTITIONS];
 };
 
 /* This copy of the table is what gets used by the encoder.