* add bench/benchVecAdd.cpp by Gael, fix crash (ei_pload on non-aligned)

* introduce packet(int), make use of it in linear vectorized paths
  --> completely fixes the slowdown noticed in benchVecAdd.
* generalize coeff(int) to linear-access xprs
* clarify the access flag bits
* rework api dox in Coeffs.h and util/Constants.h
* improve certain expressions's flags, allowing more vectorization
* fix bug in Block: start(int) and end(int) returned dyn*dyn size
* fix bug in Block: just because the Eval type has packet access
  doesn't imply the block xpr should have it too.

1 files changed, 14 insertions, 1 deletions

diff --git a/Eigen/src/Core/CwiseNullaryOp.h b/Eigen/src/Core/CwiseNullaryOp.h
index 069fad8a2..998b7ce56 100644
--- a/Eigen/src/Core/CwiseNullaryOp.h
+++ b/Eigen/src/Core/CwiseNullaryOp.h
@@ -50,7 +50,9 @@ struct ei_traits<CwiseNullaryOp<NullaryOp, MatrixType> >
     MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime,
     MaxColsAtCompileTime = MatrixType::MaxColsAtCompileTime,
     Flags = (MatrixType::Flags
-      & (HereditaryBits | LinearAccessBit | (ei_functor_traits<NullaryOp>::PacketAccess ? PacketAccessBit : 0)))
+      & (  HereditaryBits
+         | (ei_functor_has_linear_access<NullaryOp>::ret ? LinearAccessBit : 0)
+         | (ei_functor_traits<NullaryOp>::PacketAccess ? PacketAccessBit : 0)))
       | (ei_functor_traits<NullaryOp>::IsRepeatable ? 0 : EvalBeforeNestingBit),
     CoeffReadCost = ei_functor_traits<NullaryOp>::Cost
   };
@@ -89,6 +91,17 @@ class CwiseNullaryOp : ei_no_assignment_operator,
       return m_functor.packetOp();
     }
 
+    const Scalar _coeff(int index) const
+    {
+      return m_functor(index);
+    }
+
+    template<int LoadMode>
+    PacketScalar _packet(int) const
+    {
+      return m_functor.packetOp();
+    }
+
   protected:
     const ei_int_if_dynamic<RowsAtCompileTime> m_rows;
     const ei_int_if_dynamic<ColsAtCompileTime> m_cols;