diff options
author | Benoit Steiner <benoit.steiner.goog@gmail.com> | 2014-04-23 18:22:10 -0700 |
---|---|---|
committer | Benoit Steiner <benoit.steiner.goog@gmail.com> | 2014-04-23 18:22:10 -0700 |
commit | ccb4dec719c1dffbd3565497cd897abb1b9f730f (patch) | |
tree | 4a469e085ec21d31968f08e3a99e26c56d04bf2c /Eigen/src/Core/arch/NEON/Complex.h | |
parent | 82b09fcb911c5a0734ccf441504269f33034194a (diff) |
Created a NEON version of the ptranspose packet primitives
Diffstat (limited to 'Eigen/src/Core/arch/NEON/Complex.h')
-rw-r--r-- | Eigen/src/Core/arch/NEON/Complex.h | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/Eigen/src/Core/arch/NEON/Complex.h b/Eigen/src/Core/arch/NEON/Complex.h index 7b94733ab..a668382e2 100644 --- a/Eigen/src/Core/arch/NEON/Complex.h +++ b/Eigen/src/Core/arch/NEON/Complex.h @@ -263,6 +263,14 @@ template<> EIGEN_STRONG_INLINE Packet2cf pdiv<Packet2cf>(const Packet2cf& a, con return Packet2cf(pdiv(res.v, vaddq_f32(s,rev_s))); } +template<> EIGEN_DEVICE_FUNC inline void +ptranspose(Kernel<Packet2cf>& kernel) { + float32x4_t tmp = vcombine_f32(vget_high_f32(kernel.packet[0].v), vget_high_f32(kernel.packet[1].v)); + kernel.packet[0].v = vcombine_f32(vget_low_f32(kernel.packet[0].v), vget_low_f32(kernel.packet[1].v)); + kernel.packet[1].v = tmp; +} + + } // end namespace internal } // end namespace Eigen |