Extend support for Packet16b:

* Add ptranspose<*,4> to support matmul and add unit test for Matrix<bool> * Matrix<bool>
* work around a bug in slicing of Tensor<bool>.
* Add tensor tests

This speeds up matmul for boolean matrices by about 10x

name                            old time/op             new time/op             delta
BM_MatMul<bool>/8                267ns ± 0%              479ns ± 0%  +79.25%          (p=0.008 n=5+5)
BM_MatMul<bool>/32              6.42µs ± 0%             0.87µs ± 0%  -86.50%          (p=0.008 n=5+5)
BM_MatMul<bool>/64              43.3µs ± 0%              5.9µs ± 0%  -86.42%          (p=0.008 n=5+5)
BM_MatMul<bool>/128              315µs ± 0%               44µs ± 0%  -85.98%          (p=0.008 n=5+5)
BM_MatMul<bool>/256             2.41ms ± 0%             0.34ms ± 0%  -85.68%          (p=0.008 n=5+5)
BM_MatMul<bool>/512             18.8ms ± 0%              2.7ms ± 0%  -85.53%          (p=0.008 n=5+5)
BM_MatMul<bool>/1k               149ms ± 0%               22ms ± 0%  -85.40%          (p=0.008 n=5+5)

1 files changed, 21 insertions, 23 deletions

diff --git a/Eigen/src/Core/functors/BinaryFunctors.h b/Eigen/src/Core/functors/BinaryFunctors.h
index a2bc58c76..697816663 100644
--- a/Eigen/src/Core/functors/BinaryFunctors.h
+++ b/Eigen/src/Core/functors/BinaryFunctors.h
@@ -39,12 +39,12 @@ struct scalar_sum_op : binary_op_base<LhsScalar,RhsScalar>
     EIGEN_SCALAR_BINARY_OP_PLUGIN
   }
 #endif
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return a + b; }
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return a + b; }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::padd(a,b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type predux(const Packet& a) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type predux(const Packet& a) const
   { return internal::predux(a); }
 };
 template<typename LhsScalar,typename RhsScalar>
@@ -56,15 +56,9 @@ struct functor_traits<scalar_sum_op<LhsScalar,RhsScalar> > {
   };
 };
 
-/** \internal
-  * \brief Template specialization to deprecate the summation of boolean expressions.
-  * This is required to solve Bug 426.
-  * \sa DenseBase::count(), DenseBase::any(), ArrayBase::cast(), MatrixBase::cast()
-  */
-template<> struct scalar_sum_op<bool,bool> : scalar_sum_op<int,int> {
-  EIGEN_DEPRECATED
-  scalar_sum_op() {}
-};
+
+template<>
+EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE bool scalar_sum_op<bool,bool>::operator() (const bool& a, const bool& b) const { return a || b; }
 
 
 /** \internal
@@ -83,12 +77,12 @@ struct scalar_product_op  : binary_op_base<LhsScalar,RhsScalar>
     EIGEN_SCALAR_BINARY_OP_PLUGIN
   }
 #endif
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return a * b; }
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return a * b; }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::pmul(a,b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type predux(const Packet& a) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type predux(const Packet& a) const
   { return internal::predux_mul(a); }
 };
 template<typename LhsScalar,typename RhsScalar>
@@ -100,6 +94,10 @@ struct functor_traits<scalar_product_op<LhsScalar,RhsScalar> > {
   };
 };
 
+template<>
+EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE bool scalar_product_op<bool,bool>::operator() (const bool& a, const bool& b) const { return a && b; }
+
+
 /** \internal
   * \brief Template functor to compute the conjugate product of two scalars
   *
@@ -116,11 +114,11 @@ struct scalar_conj_product_op  : binary_op_base<LhsScalar,RhsScalar>
   typedef typename ScalarBinaryOpTraits<LhsScalar,RhsScalar,scalar_conj_product_op>::ReturnType result_type;
   
   EIGEN_EMPTY_STRUCT_CTOR(scalar_conj_product_op)
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type operator() (const LhsScalar& a, const RhsScalar& b) const
   { return conj_helper<LhsScalar,RhsScalar,Conj,false>().pmul(a,b); }
   
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet packetOp(const Packet& a, const Packet& b) const
   { return conj_helper<Packet,Packet,Conj,false>().pmul(a,b); }
 };
 template<typename LhsScalar,typename RhsScalar>
@@ -141,12 +139,12 @@ struct scalar_min_op : binary_op_base<LhsScalar,RhsScalar>
 {
   typedef typename ScalarBinaryOpTraits<LhsScalar,RhsScalar,scalar_min_op>::ReturnType result_type;
   EIGEN_EMPTY_STRUCT_CTOR(scalar_min_op)
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return numext::mini(a, b); }
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return numext::mini(a, b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::pmin(a,b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type predux(const Packet& a) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type predux(const Packet& a) const
   { return internal::predux_min(a); }
 };
 template<typename LhsScalar,typename RhsScalar>
@@ -167,12 +165,12 @@ struct scalar_max_op  : binary_op_base<LhsScalar,RhsScalar>
 {
   typedef typename ScalarBinaryOpTraits<LhsScalar,RhsScalar,scalar_max_op>::ReturnType result_type;
   EIGEN_EMPTY_STRUCT_CTOR(scalar_max_op)
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return numext::maxi(a, b); }
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return numext::maxi(a, b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::pmax(a,b); }
   template<typename Packet>
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const result_type predux(const Packet& a) const
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE result_type predux(const Packet& a) const
   { return internal::predux_max(a); }
 };
 template<typename LhsScalar,typename RhsScalar>