diff options
author | Gael Guennebaud <g.gael@free.fr> | 2014-04-16 17:05:11 +0200 |
---|---|---|
committer | Gael Guennebaud <g.gael@free.fr> | 2014-04-16 17:05:11 +0200 |
commit | d5a795f67366db20a132cc70e4f0217f42372357 (patch) | |
tree | 74df7a911811e64a4fa0baff940abe9c97abd5b6 /Eigen/src/Core/arch/AVX/Complex.h | |
parent | feaf7c7e6d01a4804cee5949a01ece1f8a46866f (diff) |
New gebp kernel handling up to 3 packets x 4 register-level blocks. Huge speeup on Haswell.
This changeset also introduce new vector functions: ploadquad and predux4.
Diffstat (limited to 'Eigen/src/Core/arch/AVX/Complex.h')
-rw-r--r-- | Eigen/src/Core/arch/AVX/Complex.h | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/Eigen/src/Core/arch/AVX/Complex.h b/Eigen/src/Core/arch/AVX/Complex.h index cb16180c5..8f95a7be7 100644 --- a/Eigen/src/Core/arch/AVX/Complex.h +++ b/Eigen/src/Core/arch/AVX/Complex.h @@ -45,7 +45,7 @@ template<> struct packet_traits<std::complex<float> > : default_packet_traits }; }; -template<> struct unpacket_traits<Packet4cf> { typedef std::complex<float> type; enum {size=4}; }; +template<> struct unpacket_traits<Packet4cf> { typedef std::complex<float> type; enum {size=4}; typedef Packet2cf half; }; template<> EIGEN_STRONG_INLINE Packet4cf padd<Packet4cf>(const Packet4cf& a, const Packet4cf& b) { return Packet4cf(_mm256_add_ps(a.v,b.v)); } template<> EIGEN_STRONG_INLINE Packet4cf psub<Packet4cf>(const Packet4cf& a, const Packet4cf& b) { return Packet4cf(_mm256_sub_ps(a.v,b.v)); } @@ -271,7 +271,7 @@ template<> struct packet_traits<std::complex<double> > : default_packet_traits }; }; -template<> struct unpacket_traits<Packet2cd> { typedef std::complex<double> type; enum {size=2}; }; +template<> struct unpacket_traits<Packet2cd> { typedef std::complex<double> type; enum {size=2}; typedef Packet1cd half; }; template<> EIGEN_STRONG_INLINE Packet2cd padd<Packet2cd>(const Packet2cd& a, const Packet2cd& b) { return Packet2cd(_mm256_add_pd(a.v,b.v)); } template<> EIGEN_STRONG_INLINE Packet2cd psub<Packet2cd>(const Packet2cd& a, const Packet2cd& b) { return Packet2cd(_mm256_sub_pd(a.v,b.v)); } |