we were already aligning to 16 byte boundary fixed-size objects that are multiple of 16 bytes;

now we also align to 8byte boundary fixed-size objects that are multiple of 8 bytes.
That's only useful for now for double, not e.g. for Vector2f, but that didn't seem to hurt. Am I missing something? Do you prefer that we don't align Vector2f at all?
Also, improvements in test_unalignedassert.

2 files changed, 6 insertions, 6 deletions

diff --git a/Eigen/src/Core/arch/AltiVec/PacketMath.h b/Eigen/src/Core/arch/AltiVec/PacketMath.h
index a9c16200e..1526a4b97 100644
--- a/Eigen/src/Core/arch/AltiVec/PacketMath.h
+++ b/Eigen/src/Core/arch/AltiVec/PacketMath.h
@@ -265,14 +265,14 @@ template<> inline void ei_pstoreu(int*    to , const v4i&    from )
 
 template<> inline float  ei_pfirst(const v4f&  a)
 {
-  float EIGEN_ALIGN_128 af[4];
+  float EIGEN_ALIGN16 af[4];
   vec_st(a, 0, af);
   return af[0];
 }
 
 template<> inline int    ei_pfirst(const v4i&  a)
 {
-  int EIGEN_ALIGN_128 ai[4];
+  int EIGEN_ALIGN16 ai[4];
   vec_st(a, 0, ai);
   return ai[0];
 }
@@ -373,7 +373,7 @@ inline float ei_predux_mul(const v4f& a)
 
 inline int ei_predux_mul(const v4i& a)
 {
-  EIGEN_ALIGN_128 int aux[4];
+  EIGEN_ALIGN16 int aux[4];
   ei_pstore(aux, a);
   return aux[0] * aux[1] * aux[2] * aux[3];
 }
diff --git a/Eigen/src/Core/arch/SSE/PacketMath.h b/Eigen/src/Core/arch/SSE/PacketMath.h
index ddc7b4aaf..eb1c2d311 100644
--- a/Eigen/src/Core/arch/SSE/PacketMath.h
+++ b/Eigen/src/Core/arch/SSE/PacketMath.h
@@ -359,7 +359,7 @@ template<> EIGEN_STRONG_INLINE int ei_predux_mul<Packet4i>(const Packet4i& a)
   // after some experiments, it is seems this is the fastest way to implement it
   // for GCC (eg., reusing ei_pmul is very slow !)
   // TODO try to call _mm_mul_epu32 directly
-  EIGEN_ALIGN_128 int aux[4];
+  EIGEN_ALIGN16 int aux[4];
   ei_pstore(aux, a);
   return  (aux[0] * aux[1]) * (aux[2] * aux[3]);;
 }
@@ -378,7 +378,7 @@ template<> EIGEN_STRONG_INLINE int ei_predux_min<Packet4i>(const Packet4i& a)
 {
   // after some experiments, it is seems this is the fastest way to implement it
   // for GCC (eg., it does not like using std::min after the ei_pstore !!)
-  EIGEN_ALIGN_128 int aux[4];
+  EIGEN_ALIGN16 int aux[4];
   ei_pstore(aux, a);
   register int aux0 = aux[0]<aux[1] ? aux[0] : aux[1];
   register int aux2 = aux[2]<aux[3] ? aux[2] : aux[3];
@@ -399,7 +399,7 @@ template<> EIGEN_STRONG_INLINE int ei_predux_max<Packet4i>(const Packet4i& a)
 {
   // after some experiments, it is seems this is the fastest way to implement it
   // for GCC (eg., it does not like using std::min after the ei_pstore !!)
-  EIGEN_ALIGN_128 int aux[4];
+  EIGEN_ALIGN16 int aux[4];
   ei_pstore(aux, a);
   register int aux0 = aux[0]>aux[1] ? aux[0] : aux[1];
   register int aux2 = aux[2]>aux[3] ? aux[2] : aux[3];