From 7dd894e40e439a6d1f4aed659d1375d65589cff3 Mon Sep 17 00:00:00 2001
From: Gael Guennebaud <g.gael@free.fr>
Date: Tue, 22 Nov 2016 21:41:30 +0100
Subject: Optimize predux<Packet4d> (AVX)

---
 Eigen/src/Core/arch/AVX/PacketMath.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

(limited to 'Eigen/src/Core/arch')

diff --git a/Eigen/src/Core/arch/AVX/PacketMath.h b/Eigen/src/Core/arch/AVX/PacketMath.h
index e60ef307b..190e9a42b 100644
--- a/Eigen/src/Core/arch/AVX/PacketMath.h
+++ b/Eigen/src/Core/arch/AVX/PacketMath.h
@@ -401,8 +401,7 @@ template<> EIGEN_STRONG_INLINE float predux<Packet8f>(const Packet8f& a)
 }
 template<> EIGEN_STRONG_INLINE double predux<Packet4d>(const Packet4d& a)
 {
-  Packet4d tmp0 = _mm256_hadd_pd(a,_mm256_permute2f128_pd(a,a,1));
-  return pfirst(_mm256_hadd_pd(tmp0,tmp0));
+  return predux(_mm_add_pd(_mm256_castpd256_pd128(a),_mm256_extractf128_pd(a,1)));
 }
 
 template<> EIGEN_STRONG_INLINE Packet4f predux_downto4<Packet8f>(const Packet8f& a)
-- 
cgit v1.2.3