Calculates a fast approximation for the log2 function. More...

Inheritance diagram for vctr::expressions::FastLog2< extent, SrcType >:

Public Types
using	value_type = ValueType< SrcType >

using	Constants = detail::FastLog2Constants< value_type >

Public Member Functions
VCTR_FORCEDINLINE value_type	operator[] (size_t i) const

VCTR_FORCEDINLINE	VCTR_TARGET ("avx") void prepareAVXEvaluation() const

VCTR_FORCEDINLINE	VCTR_TARGET ("fma") AVXRegister< value_type > getAVX(size_t i) const

VCTR_FORCEDINLINE	VCTR_TARGET ("sse4.1") void prepareSSEEvaluation() const

VCTR_FORCEDINLINE	VCTR_TARGET ("sse4.1") SSERegister< value_type > getSSE(size_t i) const

void	prepareNeonEvaluation () const

NeonRegister< value_type >	getNeon (size_t i) const

Detailed Description

template<size_t extent, class SrcType>
requires std::same_as<float, ValueType<SrcType>>
class vctr::expressions::FastLog2< extent, SrcType >

Calculates a fast approximation for the log2 function.

This implementation is inspired by https://github.com/romeric/fastapprox/blob/ccc534400ec3e0f67de4eafb53377334962d9db6/fastapprox/src/fastlog.h#L48. It relies on reinterpreting the bytes of a float as integer and manipulating the exponent and mantissa part.

The documentation for this class was generated from the following file:

Public Types