diff options
author | Gustavo Lima Chaves <gustavo.lima.chaves@intel.com> | 2018-12-19 14:24:44 -0800 |
---|---|---|
committer | Gustavo Lima Chaves <gustavo.lima.chaves@intel.com> | 2018-12-19 14:24:44 -0800 |
commit | e763fcd09e620300226ca22d152b94867123b603 (patch) | |
tree | 8f2546814d92e2f460e2b1abe385d96d50e1ef73 /Eigen/src/Core/arch/AVX/PacketMath.h | |
parent | efa4c9c40fbed9506aaef6a5393d27713f161984 (diff) |
Introducing "vectorized" byte on unpacket_traits structs
This is a preparation to a change on gebp_traits, where a new template
argument will be introduced to dictate the packet size, so it won't be
bound to the current/max packet size only anymore.
By having packet types defined early on gebp_traits, one has now to
act on packet types, not scalars anymore, for the enum values defined
on that class. One approach for reaching the vectorizable/size
properties one needs there could be getting the packet's scalar again
with unpacket_traits<>, then the size/Vectorizable enum entries from
packet_traits<>. It turns out guards like "#ifndef
EIGEN_VECTORIZE_AVX512" at AVX/PacketMath.h will hide smaller packet
variations of packet_traits<> for some types (and it makes sense to
keep that). In other words, one can't go back to the scalar and create
a new PacketType, as this will always lead to the maximum packet type
for the architecture.
The less costly/invasive solution for that, thus, is to add the
vectorizable info on every unpacket_traits struct as well.
Diffstat (limited to 'Eigen/src/Core/arch/AVX/PacketMath.h')
-rw-r--r-- | Eigen/src/Core/arch/AVX/PacketMath.h | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/Eigen/src/Core/arch/AVX/PacketMath.h b/Eigen/src/Core/arch/AVX/PacketMath.h index e5aeb6375..e771c0f25 100644 --- a/Eigen/src/Core/arch/AVX/PacketMath.h +++ b/Eigen/src/Core/arch/AVX/PacketMath.h @@ -117,14 +117,14 @@ template<> struct unpacket_traits<Packet8f> { typedef float type; typedef Packet4f half; typedef Packet8i integer_packet; - enum {size=8, alignment=Aligned32}; + enum {size=8, alignment=Aligned32, vectorizable=true}; }; template<> struct unpacket_traits<Packet4d> { typedef double type; typedef Packet2d half; - enum {size=4, alignment=Aligned32}; + enum {size=4, alignment=Aligned32, vectorizable=true}; }; -template<> struct unpacket_traits<Packet8i> { typedef int type; typedef Packet4i half; enum {size=8, alignment=Aligned32}; }; +template<> struct unpacket_traits<Packet8i> { typedef int type; typedef Packet4i half; enum {size=8, alignment=Aligned32, vectorizable=false}; }; template<> EIGEN_STRONG_INLINE Packet8f pset1<Packet8f>(const float& from) { return _mm256_set1_ps(from); } template<> EIGEN_STRONG_INLINE Packet4d pset1<Packet4d>(const double& from) { return _mm256_set1_pd(from); } |