devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix ldexp implementations.

Browse Source 
The previous implementations produced garbage values if the exponent did
not fit within the exponent bits.  See #2131 for a complete discussion,
and !375 for other possible implementations.

Here we implement the 4-factor version. See `pldexp_impl` in
`GenericPacketMathFunctions.h` for a full description.

The SSE `pcmp*` methods were moved down since `pcmp_le<Packet4i>`
requires `por`.

Left as a "TODO" is to delegate to a faster version if we know the
exponent does fit within the exponent bits.

Fixes #2131.
This commit is contained in:
Antonio Sanchez
2020-10-12 12:24:08 +01:00
committed by Rasmus Munk Larsen
parent 7eb07da538
commit 4cb563a01e
8 changed files with 354 additions and 112 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
Eigen/src/Core/GenericPacketMath.h
View File

@@ -453,7 +453,7 @@ EIGEN_DEVICE_FUNC inline Packet pfrexp(const Packet& a, Packet& exponent) {
return result;
}
/** \internal \returns a * 2^exponent
/** \internal \returns a * 2^((int)exponent)
* See https://en.cppreference.com/w/cpp/numeric/math/ldexp
*/
template<typename Packet> EIGEN_DEVICE_FUNC inline Packet
@@ -905,28 +905,6 @@ pblend(const Selector<unpacket_traits<Packet>::size>& ifPacket, const Packet& th
return ifPacket.select[0] ? thenPacket : elsePacket;
}
/***************************************************************************
* Some generic implementations to be used by implementors
***************************************************************************/
/** Default implementation of pfrexp for float.
* It is expected to be called by implementers of template<> pfrexp.
*/
template<typename Packet> EIGEN_STRONG_INLINE Packet
pfrexp_float(const Packet& a, Packet& exponent);
/** Default implementation of pldexp for float.
* It is expected to be called by implementers of template<> pldexp.
*/
template<typename Packet> EIGEN_STRONG_INLINE Packet
pldexp_float(Packet a, Packet exponent);
/** Default implementation of pldexp for double.
* It is expected to be called by implementers of template<> pldexp.
*/
template<typename Packet> EIGEN_STRONG_INLINE Packet
pldexp_double(Packet a, Packet exponent);
} // end namespace internal
} // end namespace Eigen