poplar::IeeeHalf Class Reference

A IeeeHalf. More...

#include <IeeeHalf.hpp>

Public Member Functions
	IeeeHalf ()=default
	Uninitialised.
	IeeeHalf (float value)
	Initialise from a single-precision fp value.
	operator float () const
	Type-cast to single-precision.
uint16_t	bit16 () const
	Obtain half-precision bit-pattern. More...
uint32_t	bitz32 () const
	Obtain half-precision bit-pattern. More...

Static Public Member Functions
static IeeeHalf	fromBits (uint16_t bitPattern)
	Initialise from a raw 16-bit pattern (conforming to IEEE 754-2008 binary16 format)

Detailed Description

A IeeeHalf.

The half is stored in memory as an actual IeeeHalf but most operations are performed by converting it to a float, performing the operation and converting back again.

Member Function Documentation

bit16()

uint16_t poplar::IeeeHalf::bit16 ( ) const

inline

Obtain half-precision bit-pattern.

Returns

raw 16-bit bit-pattern, as described by IEEE 754-2008

bitz32()

uint32_t poplar::IeeeHalf::bitz32 ( ) const

inline

Obtain half-precision bit-pattern.

Returns

raw zero-extended 16-bit bit-pattern, as described by IEEE 754-2008

---

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

• include/runtime/IeeeHalf.hpp