FFloat16 Class Reference

#include <Float16.h>

Public Member Functions
	FFloat16 ()=default
	FFloat16 (const FFloat16 &FP16Value)=default
	FFloat16 (float FP32Value)
FFloat16 &	operator= (float FP32Value)
FFloat16 &	operator= (const FFloat16 &FP16Value)=default
	operator float () const
void	Set (float FP32Value)
void	SetClamped (float FP32Value)
void CORE_API	SetTruncate (float FP32Value)
void	SetZero ()
void	SetOne ()
FFloat16	GetClampedNonNegativeAndFinite () const
FFloat16	GetClampedFinite () const
float	GetFloat () const
bool	IsNegative () const

Public Attributes
uint16	Encoded = 0

Static Public Attributes
static constexpr float	MaxF16Float = 65504.f

Friends
FArchive &	operator<< (FArchive &Ar, FFloat16 &V)

Detailed Description

16 bit float components and conversion

IEEE float 16 Represented by 10-bit mantissa M, 5-bit exponent E, and 1-bit sign S

Specials:

E=0, M=0 == 0.0 E=0, M!=0 == Denormalized value (M / 2^10) * 2^-14 0<E<31, M=any == (1 + M / 2^10) * 2^(E-15) E=31, M=0 == Infinity E=31, M!=0 == NAN

conversion from 32 bit float is with RTNE (round to nearest even)

Legacy code truncated in the conversion. SetTruncate can be used for backwards compatibility.

Constructor & Destructor Documentation

FFloat16() [1/3]

FFloat16::FFloat16 ( )

default

Default constructor

FFloat16() [2/3]

FFloat16::FFloat16 ( const FFloat16 &  FP16Value )

default

Copy constructor.

FFloat16() [3/3]

UE_FORCEINLINE_HINT FFloat16::FFloat16

(

float

FP32Value

)

Conversion constructor. Convert from Fp32 to Fp16.

Member Function Documentation

GetClampedFinite()

FFloat16 FFloat16::GetClampedFinite ( ) const

inline

Return float clamp in [-MaxF16Float,MaxF16Float] , no infinites or nans returned

GetClampedNonNegativeAndFinite()

FFloat16 FFloat16::GetClampedNonNegativeAndFinite ( ) const

inline

Return float clamp in [0,MaxF16Float] , no negatives or infinites or nans returned

GetFloat()

UE_FORCEINLINE_HINT float FFloat16::GetFloat

(

)

const

Convert from Fp16 to Fp32.

IsNegative()

bool FFloat16::IsNegative ( ) const

inline

Is the float negative without converting NOTE: returns true for negative zero!

operator float()

UE_FORCEINLINE_HINT FFloat16::operator float

(

)

const

Convert from Fp16 to Fp32.

operator=() [1/2]

FFloat16 & FFloat16::operator= ( const FFloat16 &  FP16Value )

default

Assignment operator. Copy Fp16 value.

operator=() [2/2]

FFloat16 & FFloat16::operator= ( float  FP32Value )

inline

Assignment operator. Convert from Fp32 to Fp16.

Set()

UE_FORCEINLINE_HINT void FFloat16::Set

(

float

FP32Value

)

Convert from Fp32 to Fp16, round-to-nearest-even. (RTNE) Stores values out of range as +-Inf

SetClamped()

void FFloat16::SetClamped ( float  FP32Value )

inline

SetOne()

void FFloat16::SetOne ( )

inline

Set to 1.0

SetTruncate()

void FFloat16::SetTruncate

(

float

FP32Value

)

Convert from Fp32 to Fp16, truncating low bits. (backward-compatible conversion; was used by Set() previously) Clamps values out of range to [-MaxF16Float,MaxF16Float]

SetZero()

void FFloat16::SetZero ( )

inline

Set to 0.0

Friends And Related Symbol Documentation

operator<<

FArchive & operator<< ( FArchive &  Ar, FFloat16 &  V  )

friend

Serializes the FFloat16.

Parameters

Ar	Reference to the serialization archive.
V	Reference to the FFloat16 being serialized.

Returns

Reference to the Archive after serialization.

Member Data Documentation

Encoded

uint16 FFloat16::Encoded = 0

MaxF16Float

constexpr float FFloat16::MaxF16Float = 65504.f

staticconstexpr

---

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

• Engine/Source/Runtime/Core/Public/Math/Float16.h
• Engine/Source/Runtime/Core/Private/Math/Float16.cpp