Data alignment fix for SSIM.

WebRtc_UWord64[2] wasn't always aligned to 128 bytes, which
is necessary for _mm_store_si128.  By declaring the 
variable as __m128i it will always be 128 bytes aligned.

Incorrect include files.

__m128i is defined in emmintrin.h for visual studio.  Extra include on mac and linux is not a problem.
Review URL: http://webrtc-codereview.appspot.com/239013

git-svn-id: http://webrtc.googlecode.com/svn/trunk@816 4adac7df-926f-26a2-2b94-8c16560cd09d

Author: [email protected]

src/modules/video_coding/main/test/video_metrics.cc

@@ -164,6 +164,7 @@ Ssim8x8C(WebRtc_UWord8 *s, WebRtc_Word32 sp,
 }
 
 #if defined(WEBRTC_USE_SSE2)
+#include <emmintrin.h>
 #include <xmmintrin.h>
 static double
 Ssim8x8Sse2(WebRtc_UWord8 *s, WebRtc_Word32 sp,
@@ -200,28 +201,39 @@ Ssim8x8Sse2(WebRtc_UWord8 *s, WebRtc_Word32 sp,
     const __m128i sum_r_32  = _mm_add_epi32(_mm_unpackhi_epi16(sum_r_16, z),
                                             _mm_unpacklo_epi16(sum_r_16, z));
 
-    WebRtc_UWord64 sum_s_64[2];
-    WebRtc_UWord64 sum_r_64[2];
-    WebRtc_UWord64 sum_sq_s_64[2];
-    WebRtc_UWord64 sum_sq_r_64[2];
-    WebRtc_UWord64 sum_sxr_64[2];
+    __m128i sum_s_128;
+    __m128i sum_r_128;
+    __m128i sum_sq_s_128;
+    __m128i sum_sq_r_128;
+    __m128i sum_sxr_128;
 
-    _mm_store_si128 ((__m128i*)sum_s_64,
+    _mm_store_si128 (&sum_s_128,
                       _mm_add_epi64(_mm_unpackhi_epi32(sum_s_32, z),
                                     _mm_unpacklo_epi32(sum_s_32, z)));
-    _mm_store_si128 ((__m128i*)sum_r_64,
+    _mm_store_si128 (&sum_r_128,
                       _mm_add_epi64(_mm_unpackhi_epi32(sum_r_32, z),
                                     _mm_unpacklo_epi32(sum_r_32, z)));
-    _mm_store_si128 ((__m128i*)sum_sq_s_64,
+    _mm_store_si128 (&sum_sq_s_128,
                       _mm_add_epi64(_mm_unpackhi_epi32(sum_sq_s_32, z),
                                     _mm_unpacklo_epi32(sum_sq_s_32, z)));
-    _mm_store_si128 ((__m128i*)sum_sq_r_64,
+    _mm_store_si128 (&sum_sq_r_128,
                       _mm_add_epi64(_mm_unpackhi_epi32(sum_sq_r_32, z),
                                     _mm_unpacklo_epi32(sum_sq_r_32, z)));
-    _mm_store_si128 ((__m128i*)sum_sxr_64,
+    _mm_store_si128 (&sum_sxr_128,
                       _mm_add_epi64(_mm_unpackhi_epi32(sum_sxr_32, z),
                                     _mm_unpacklo_epi32(sum_sxr_32, z)));
 
+    const WebRtc_UWord64 *sum_s_64 =
+                          reinterpret_cast<WebRtc_UWord64*>(&sum_s_128);
+    const WebRtc_UWord64 *sum_r_64 =
+                          reinterpret_cast<WebRtc_UWord64*>(&sum_r_128);
+    const WebRtc_UWord64 *sum_sq_s_64 =
+                          reinterpret_cast<WebRtc_UWord64*>(&sum_sq_s_128);
+    const WebRtc_UWord64 *sum_sq_r_64 =
+                          reinterpret_cast<WebRtc_UWord64*>(&sum_sq_r_128);
+    const WebRtc_UWord64 *sum_sxr_64 =
+                          reinterpret_cast<WebRtc_UWord64*>(&sum_sxr_128);
+
     const WebRtc_UWord64 sum_s    = sum_s_64[0]    + sum_s_64[1];
     const WebRtc_UWord64 sum_r    = sum_r_64[0]    + sum_r_64[1];
     const WebRtc_UWord64 sum_sq_s = sum_sq_s_64[0] + sum_sq_s_64[1];