[x265] [PATCH 1 of 1] sse.inc: Modified sse_ss8 intrinsic function. Removed redundancy using comman macro PROCESS_SSE_SS4x1

[dnyaneshwar at multicorewareinc.com]

# HG changeset patch
# User Dnyaneshwar Gorade <dnyaneshwar at multicorewareinc.com>
# Date 1381324279 -19800
#      Wed Oct 09 18:41:19 2013 +0530
# Node ID bf5e720d83f4c7eab13c19df7a8e159086a8f728
# Parent  dcd7432686e2c9cbe41aa9d9f74482e1d3f1ab47
sse.inc: Modified sse_ss8 intrinsic function. Removed redundancy using comman macro PROCESS_SSE_SS4x1.

diff -r dcd7432686e2 -r bf5e720d83f4 source/common/vec/sse.inc
--- a/source/common/vec/sse.inc	Wed Oct 09 18:34:53 2013 +0530
+++ b/source/common/vec/sse.inc	Wed Oct 09 18:41:19 2013 +0530
@@ -82,26 +82,24 @@
 int sse_ss8(short* fenc, intptr_t strideFenc, short* fref, intptr_t strideFref)
 {
     int rows = ly;
+    __m128i sum  = _mm_setzero_si128();
+    __m128i m1, n1, diff, sign1, sign2, tmp1, tmp2;
 
-    Vec4i diff(0);
-    Vec8s m1, n1;
-    Vec4i sum(0);
     for (; rows != 0; rows--)
     {
-        m1.load(fenc);
-        n1.load(fref);
-        diff = extend_low(m1) - extend_low(n1);
-        diff = diff * diff;
-        sum += diff;
+        PROCESS_SSE_SS4x1(0);
 
-        diff = extend_high(m1) - extend_high(n1);
-        diff = diff * diff;
-        sum += diff;
+        m1 = _mm_unpackhi_epi16(m1, sign1);
+        n1 = _mm_unpackhi_epi16(n1, sign2);
+        diff = _mm_sub_epi32(m1, n1);
+        diff = _mm_mullo_epi32(diff, diff);
+        sum = _mm_add_epi32(sum, diff);
+
         fenc += strideFenc;
         fref += strideFref;
     }
-
-    return horizontal_add(sum);
+    __m128i sum1  = _mm_hadd_epi32(sum, sum);
+    return _mm_cvtsi128_si32(_mm_hadd_epi32(sum1, sum1));
 }
 
 template<int ly>