diff options
Diffstat (limited to 'Eigen/src/Core/arch/AVX512/PacketMath.h')
| -rw-r--r-- | Eigen/src/Core/arch/AVX512/PacketMath.h | 13 |
1 files changed, 10 insertions, 3 deletions
diff --git a/Eigen/src/Core/arch/AVX512/PacketMath.h b/Eigen/src/Core/arch/AVX512/PacketMath.h index 60b723b08..094309eda 100644 --- a/Eigen/src/Core/arch/AVX512/PacketMath.h +++ b/Eigen/src/Core/arch/AVX512/PacketMath.h @@ -102,19 +102,20 @@ struct unpacket_traits<Packet16f> { typedef float type; typedef Packet8f half; typedef Packet16i integer_packet; - enum { size = 16, alignment=Aligned64, vectorizable=true }; + typedef uint16_t mask_t; + enum { size = 16, alignment=Aligned64, vectorizable=true, masked_load_available=true }; }; template <> struct unpacket_traits<Packet8d> { typedef double type; typedef Packet4d half; - enum { size = 8, alignment=Aligned64, vectorizable=true }; + enum { size = 8, alignment=Aligned64, vectorizable=true, masked_load_available=false }; }; template <> struct unpacket_traits<Packet16i> { typedef int type; typedef Packet8i half; - enum { size = 16, alignment=Aligned64, vectorizable=false }; + enum { size = 16, alignment=Aligned64, vectorizable=false, masked_load_available=false }; }; template <> @@ -485,6 +486,12 @@ EIGEN_STRONG_INLINE Packet16i ploadu<Packet16i>(const int* from) { reinterpret_cast<const __m512i*>(from)); } +template <> +EIGEN_STRONG_INLINE Packet16f ploadu<Packet16f>(const float* from, uint16_t umask) { + __mmask16 mask = static_cast<__mmask16>(umask); + EIGEN_DEBUG_UNALIGNED_LOAD return _mm512_maskz_loadu_ps(mask, from); +} + // Loads 8 floats from memory a returns the packet // {a0, a0 a1, a1, a2, a2, a3, a3, a4, a4, a5, a5, a6, a6, a7, a7} template <> |