diff --git a/README.chromium b/README.chromium
index 252614510..f4aa928b8 100644
--- a/README.chromium
+++ b/README.chromium
@@ -1,6 +1,6 @@
 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 1615
+Version: 1616
 License: BSD
 License File: LICENSE
 
diff --git a/include/libyuv/row.h b/include/libyuv/row.h
index f7b09b108..013a7e53e 100644
--- a/include/libyuv/row.h
+++ b/include/libyuv/row.h
@@ -373,8 +373,11 @@ extern "C" {
 #endif
 
 #if defined(_MSC_VER) && !defined(__CLR_VER) && !defined(__clang__)
+#if defined(VISUALC_HAS_AVX2)
+#define SIMD_ALIGNED(var) __declspec(align(32)) var
+#else
 #define SIMD_ALIGNED(var) __declspec(align(16)) var
-#define SIMD_ALIGNED32(var) __declspec(align(32)) var
+#endif
 typedef __declspec(align(16)) int16 vec16[8];
 typedef __declspec(align(16)) int32 vec32[4];
 typedef __declspec(align(16)) int8 vec8[16];
@@ -389,8 +392,11 @@ typedef __declspec(align(32)) uint32 ulvec32[8];
 typedef __declspec(align(32)) uint8 ulvec8[32];
 #elif !defined(__pnacl__) && (defined(__GNUC__) || defined(__clang__))
 // Caveat GCC 4.2 to 4.7 have a known issue using vectors with const.
+#if defined(CLANG_HAS_AVX2) || defined(GCC_HAS_AVX2)
+#define SIMD_ALIGNED(var) var __attribute__((aligned(32)))
+#else
 #define SIMD_ALIGNED(var) var __attribute__((aligned(16)))
-#define SIMD_ALIGNED32(var) var __attribute__((aligned(32)))
+#endif
 typedef int16 __attribute__((vector_size(16))) vec16;
 typedef int32 __attribute__((vector_size(16))) vec32;
 typedef int8 __attribute__((vector_size(16))) vec8;
@@ -405,7 +411,6 @@ typedef uint32 __attribute__((vector_size(32))) ulvec32;
 typedef uint8 __attribute__((vector_size(32))) ulvec8;
 #else
 #define SIMD_ALIGNED(var) var
-#define SIMD_ALIGNED32(var) var
 typedef int16 vec16[8];
 typedef int32 vec32[4];
 typedef int8 vec8[16];
@@ -461,14 +466,14 @@ struct YuvConstants {
 #endif
 
 // Conversion matrix for YUV to RGB
-extern const struct YuvConstants kYuvI601Constants;  // BT.601
-extern const struct YuvConstants kYuvJPEGConstants;  // JPeg color space
-extern const struct YuvConstants kYuvH709Constants;  // BT.709
+extern const struct YuvConstants SIMD_ALIGNED(kYuvI601Constants);  // BT.601
+extern const struct YuvConstants SIMD_ALIGNED(kYuvJPEGConstants);  // JPeg
+extern const struct YuvConstants SIMD_ALIGNED(kYuvH709Constants);  // BT.709
 
 // Conversion matrix for YVU to BGR
-extern const struct YuvConstants kYvuI601Constants;  // BT.601
-extern const struct YuvConstants kYvuJPEGConstants;  // JPeg color space
-extern const struct YuvConstants kYvuH709Constants;  // BT.709
+extern const struct YuvConstants SIMD_ALIGNED(kYvuI601Constants);  // BT.601
+extern const struct YuvConstants SIMD_ALIGNED(kYvuJPEGConstants);  // JPeg
+extern const struct YuvConstants SIMD_ALIGNED(kYvuH709Constants);  // BT.709
 
 #if defined(__APPLE__) || defined(__x86_64__) || defined(__llvm__)
 #define OMITFP
diff --git a/include/libyuv/version.h b/include/libyuv/version.h
index 18b481c2d..0fbdc022d 100644
--- a/include/libyuv/version.h
+++ b/include/libyuv/version.h
@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 1615
+#define LIBYUV_VERSION 1616
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT
diff --git a/source/row_common.cc b/source/row_common.cc
index 229da9752..aefa38c49 100644
--- a/source/row_common.cc
+++ b/source/row_common.cc
@@ -1019,7 +1019,7 @@ const struct YuvConstants SIMD_ALIGNED(kYvuI601Constants) = {
   { 0x0101 * YG, 0, 0, 0 }
 };
 #else
-const struct YuvConstants SIMD_ALIGNED32(kYuvI601Constants) = {
+const struct YuvConstants SIMD_ALIGNED(kYuvI601Constants) = {
   { UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0,
     UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0 },
   { UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG,
@@ -1031,7 +1031,7 @@ const struct YuvConstants SIMD_ALIGNED32(kYuvI601Constants) = {
   { BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR },
   { YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG }
 };
-const struct YuvConstants SIMD_ALIGNED32(kYvuI601Constants) = {
+const struct YuvConstants SIMD_ALIGNED(kYvuI601Constants) = {
   { VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0,
     VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0 },
   { VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG,
@@ -1106,7 +1106,7 @@ const struct YuvConstants SIMD_ALIGNED(kYvuJPEGConstants) = {
   { 0x0101 * YG, 0, 0, 0 }
 };
 #else
-const struct YuvConstants SIMD_ALIGNED32(kYuvJPEGConstants) = {
+const struct YuvConstants SIMD_ALIGNED(kYuvJPEGConstants) = {
   { UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0,
     UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0 },
   { UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG,
@@ -1118,7 +1118,7 @@ const struct YuvConstants SIMD_ALIGNED32(kYuvJPEGConstants) = {
   { BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR },
   { YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG }
 };
-const struct YuvConstants SIMD_ALIGNED32(kYvuJPEGConstants) = {
+const struct YuvConstants SIMD_ALIGNED(kYvuJPEGConstants) = {
   { VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0,
     VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0 },
   { VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG,
@@ -1194,7 +1194,7 @@ const struct YuvConstants SIMD_ALIGNED(kYvuH709Constants) = {
   { 0x0101 * YG, 0, 0, 0 }
 };
 #else
-const struct YuvConstants SIMD_ALIGNED32(kYuvH709Constants) = {
+const struct YuvConstants SIMD_ALIGNED(kYuvH709Constants) = {
   { UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0,
     UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0, UB, 0 },
   { UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG,
@@ -1206,7 +1206,7 @@ const struct YuvConstants SIMD_ALIGNED32(kYuvH709Constants) = {
   { BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR },
   { YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG }
 };
-const struct YuvConstants SIMD_ALIGNED32(kYvuH709Constants) = {
+const struct YuvConstants SIMD_ALIGNED(kYvuH709Constants) = {
   { VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0,
     VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0, VR, 0 },
   { VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG,
@@ -2504,7 +2504,7 @@ void I422ToRGB565Row_AVX2(const uint8* src_y,
                           uint8* dst_rgb565,
                           const struct YuvConstants* yuvconstants,
                           int width) {
-  SIMD_ALIGNED32(uint8 row[MAXTWIDTH * 4]);
+  SIMD_ALIGNED(uint8 row[MAXTWIDTH * 4]);
   while (width > 0) {
     int twidth = width > MAXTWIDTH ? MAXTWIDTH : width;
     I422ToARGBRow_AVX2(src_y, src_u, src_v, row, yuvconstants, twidth);
@@ -2530,7 +2530,7 @@ void I422ToARGB1555Row_AVX2(const uint8* src_y,
                             const struct YuvConstants* yuvconstants,
                             int width) {
   // Row buffer for intermediate ARGB pixels.
-  SIMD_ALIGNED32(uint8 row[MAXTWIDTH * 4]);
+  SIMD_ALIGNED(uint8 row[MAXTWIDTH * 4]);
   while (width > 0) {
     int twidth = width > MAXTWIDTH ? MAXTWIDTH : width;
     I422ToARGBRow_AVX2(src_y, src_u, src_v, row, yuvconstants, twidth);
@@ -2556,7 +2556,7 @@ void I422ToARGB4444Row_AVX2(const uint8* src_y,
                             const struct YuvConstants* yuvconstants,
                             int width) {
   // Row buffer for intermediate ARGB pixels.
-  SIMD_ALIGNED32(uint8 row[MAXTWIDTH * 4]);
+  SIMD_ALIGNED(uint8 row[MAXTWIDTH * 4]);
   while (width > 0) {
     int twidth = width > MAXTWIDTH ? MAXTWIDTH : width;
     I422ToARGBRow_AVX2(src_y, src_u, src_v, row, yuvconstants, twidth);
@@ -2582,7 +2582,7 @@ void I422ToRGB24Row_AVX2(const uint8* src_y,
                             const struct YuvConstants* yuvconstants,
                             int width) {
   // Row buffer for intermediate ARGB pixels.
-  SIMD_ALIGNED32(uint8 row[MAXTWIDTH * 4]);
+  SIMD_ALIGNED(uint8 row[MAXTWIDTH * 4]);
   while (width > 0) {
     int twidth = width > MAXTWIDTH ? MAXTWIDTH : width;
     I422ToARGBRow_AVX2(src_y, src_u, src_v, row, yuvconstants, twidth);
@@ -2604,7 +2604,7 @@ void NV12ToRGB565Row_AVX2(const uint8* src_y,
                           const struct YuvConstants* yuvconstants,
                           int width) {
   // Row buffer for intermediate ARGB pixels.
-  SIMD_ALIGNED32(uint8 row[MAXTWIDTH * 4]);
+  SIMD_ALIGNED(uint8 row[MAXTWIDTH * 4]);
   while (width > 0) {
     int twidth = width > MAXTWIDTH ? MAXTWIDTH : width;
     NV12ToARGBRow_AVX2(src_y, src_uv, row, yuvconstants, twidth);