From 2bda1b0d93fb627d0c500ec48b20302d44c32cb7 Mon Sep 17 00:00:00 2001
From: Benoit Steiner <benoit.steiner.goog@gmail.com>
Date: Wed, 28 Sep 2016 17:08:41 -0700
Subject: Updated the tensor sum and mean reducer to enable them to process
 complex numbers on cuda gpus.

---
 unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h')

diff --git a/unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h b/unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h
index 760074622..eddb86597 100644
--- a/unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h
+++ b/unsupported/Eigen/CXX11/src/Tensor/TensorFunctors.h
@@ -99,7 +99,8 @@ template <typename T> struct SumReducer
   static const bool IsStateful = false;
 
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void reduce(const T t, T* accum) const {
-    (*accum) += t;
+    internal::scalar_sum_op<T> sum_op;
+    *accum = sum_op(*accum, t);
   }
   template <typename Packet>
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void reducePacket(const Packet& p, Packet* accum) const {
@@ -145,7 +146,8 @@ template <typename T> struct MeanReducer
   MeanReducer() : scalarCount_(0), packetCount_(0) { }
 
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void reduce(const T t, T* accum) {
-    (*accum) += t;
+    internal::scalar_sum_op<T> sum_op;
+    *accum = sum_op(*accum, t);
     scalarCount_++;
   }
   template <typename Packet>
-- 
cgit v1.2.3