\keyword 16-bit Floating Point Support\inmodule QtCore \inheaderfile QFloat16 More...

#include <qfloat16.h>

Collaboration diagram for qfloat16:

Public Types
using	NativeType = void

using	NearestFloat = std::conditional_t< IsNative, NativeType, float >

Public Member Functions
constexpr	qfloat16 () noexcept

	qfloat16 (Qt::Initialization) noexcept

	qfloat16 (float f) noexcept

	operator float () const noexcept

template<typename T , typename = std::enable_if_t<std::is_arithmetic_v<T> && !std::is_same_v<T, NearestFloat>>>
constexpr	qfloat16 (T value) noexcept

bool	isInf () const noexcept

bool	isNaN () const noexcept

bool	isFinite () const noexcept

Q_CORE_EXPORT int	fpClassify () const noexcept

qfloat16	copySign (qfloat16 sign) const noexcept

constexpr bool	isNormal () const noexcept

Static Public Member Functions
static constexpr qfloat16	_limit_epsilon () noexcept

static constexpr qfloat16	_limit_min () noexcept

static constexpr qfloat16	_limit_denorm_min () noexcept

static constexpr qfloat16	_limit_max () noexcept

static constexpr qfloat16	_limit_lowest () noexcept

static constexpr qfloat16	_limit_infinity () noexcept

static constexpr qfloat16	_limit_quiet_NaN () noexcept

Static Public Attributes
static constexpr bool	IsNative = QFLOAT16_IS_NATIVE

Friends
bool	qIsNull (qfloat16 f) noexcept

qfloat16	operator- (qfloat16 a) noexcept

qfloat16	operator+ (qfloat16 a, qfloat16 b) noexcept

qfloat16	operator- (qfloat16 a, qfloat16 b) noexcept

qfloat16	operator* (qfloat16 a, qfloat16 b) noexcept

qfloat16	operator/ (qfloat16 a, qfloat16 b) noexcept

QT_WARNING_PUSH QT_WARNING_DISABLE_FLOAT_COMPARE friend bool	operator> (qfloat16 a, qfloat16 b) noexcept

bool	operator< (qfloat16 a, qfloat16 b) noexcept

bool	operator>= (qfloat16 a, qfloat16 b) noexcept

bool	operator<= (qfloat16 a, qfloat16 b) noexcept

bool	operator== (qfloat16 a, qfloat16 b) noexcept

bool	operator!= (qfloat16 a, qfloat16 b) noexcept

QT_WARNING_POP friend Q_CORE_EXPORT QDataStream &	operator<< (QDataStream &ds, qfloat16 f)

Q_CORE_EXPORT QDataStream &	operator>> (QDataStream &ds, qfloat16 &f)

Q_CORE_EXPORT QTextStream &	operator<< (QTextStream &ts, qfloat16 f)

Q_CORE_EXPORT QTextStream &	operator>> (QTextStream &ts, qfloat16 &f)

Related Symbols
(Note that these are not member symbols.)
bool	qIsInfqIsInf (qfloat16 f)(float)
	Returns true if the `qfloat16` {f} is equivalent to infinity.

bool	qIsNaNqIsNaN (qfloat16 f)(float)
	Returns true if the `qfloat16` {f} is not a number (NaN).

bool	qIsFiniteqIsFinite (qfloat16 f)(float)
	Returns true if the `qfloat16` {f} is a finite number.

int	qFpClassifyqFpClassify (qfloat16 val)(float)

int	qRoundqRound (qfloat16 value)(float)
	Rounds value to the nearest integer.

qint64	qRound64qRound64 (qfloat16 value)(float)
	Rounds value to the nearest 64-bit integer.

bool	qFuzzyCompareqFuzzyCompare (qfloat16 p1, qfloat16 p2)(float, float)
	Compares the floating point value p1 and p2 and returns `true` if they are considered equal, otherwise `false`.

Q_CORE_EXPORT void	qFloatToFloat16 (qfloat16 out, const float in, qsizetype len) noexcept

Q_CORE_EXPORT void	qFloatFromFloat16 (float out, const qfloat16 in, qsizetype len) noexcept

Detailed Description

\keyword 16-bit Floating Point Support

\inmodule QtCore \inheaderfile QFloat16

Provides 16-bit floating point support.

The qfloat16 class provides support for half-precision (16-bit) floating point data. It is fully compliant with IEEE 754 as a storage type. This implies that any arithmetic operation on a qfloat16 instance results in the value first being converted to a float. This conversion to and from float is performed by hardware when possible, but on processors that do not natively support half-precision, the conversion is performed through a sequence of lookup table operations.

qfloat16 should be treated as if it were a POD (plain old data) type. Consequently, none of the supported operations need any elaboration beyond stating that it supports all arithmetic operators incident to floating point types.

Note: On x86 and x86-64 that to get hardware accelerated conversions you must compile with F16C or AVX2 enabled, or use qFloatToFloat16() and qFloatFromFloat16() which will detect F16C at runtime.

Since: 5.9

Definition at line 45 of file qfloat16.h.

Member Typedef Documentation

◆ NativeType

using qfloat16::NativeType = void

Definition at line 72 of file qfloat16.h.

◆ NearestFloat

using qfloat16::NearestFloat = std::conditional_t<IsNative, NativeType, float>

Definition at line 75 of file qfloat16.h.

Constructor & Destructor Documentation

◆ qfloat16() [1/4]

constexpr qfloat16::qfloat16 ( )

inlineconstexprnoexcept

Definition at line 77 of file qfloat16.h.

Referenced by _limit_denorm_min(), _limit_epsilon(), _limit_infinity(), _limit_lowest(), _limit_max(), _limit_min(), _limit_quiet_NaN(), and copySign().

Here is the caller graph for this function:

◆ qfloat16() [2/4]

qfloat16::qfloat16 ( Qt::Initialization )

inlineexplicitnoexcept

Since: 6.1

Constructs a qfloat16 without initializing the value.

Definition at line 78 of file qfloat16.h.

◆ qfloat16() [3/4]

QT_WARNING_PUSH qfloat16::qfloat16 ( float f )

inlinenoexcept

Definition at line 320 of file qfloat16.h.

References base, qMax(), and shift().

Here is the call graph for this function:

◆ qfloat16() [4/4]

template<typename T , typename = std::enable_if_t<std::is_arithmetic_v<T> && !std::is_same_v<T, NearestFloat>>>

constexpr qfloat16::qfloat16 ( T value )

inlineexplicitconstexprnoexcept

Definition at line 88 of file qfloat16.h.

Member Function Documentation

◆ _limit_denorm_min()

static constexpr qfloat16 qfloat16::_limit_denorm_min ( )

inlinestaticconstexprnoexcept

Definition at line 101 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_epsilon()

static constexpr qfloat16 qfloat16::_limit_epsilon ( )

inlinestaticconstexprnoexcept

Definition at line 99 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_infinity()

static constexpr qfloat16 qfloat16::_limit_infinity ( )

inlinestaticconstexprnoexcept

Definition at line 104 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_lowest()

static constexpr qfloat16 qfloat16::_limit_lowest ( )

inlinestaticconstexprnoexcept

Definition at line 103 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_max()

static constexpr qfloat16 qfloat16::_limit_max ( )

inlinestaticconstexprnoexcept

Definition at line 102 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_min()

static constexpr qfloat16 qfloat16::_limit_min ( )

inlinestaticconstexprnoexcept

Definition at line 100 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ _limit_quiet_NaN()

static constexpr qfloat16 qfloat16::_limit_quiet_NaN ( )

inlinestaticconstexprnoexcept

Definition at line 105 of file qfloat16.h.

References qfloat16().

Here is the call graph for this function:

◆ copySign()

qfloat16 qfloat16::copySign ( qfloat16 sign ) const

inlinenoexcept

Since: 5.15

Returns a qfloat16 with the sign of sign but the rest of its value taken from this qfloat16. Serves as qfloat16's equivalent of std::copysign().

Definition at line 96 of file qfloat16.h.

References qfloat16(), b16, and sign().

Here is the call graph for this function:

◆ fpClassify()

int qfloat16::fpClassify ( ) const

noexcept

Since: 5.14 Implements qFpClassify() for qfloat16.

Definition at line 132 of file qfloat16.cpp.

◆ isFinite()

bool qfloat16::isFinite ( ) const

inlinenoexcept

Since: 5.14

Tests whether this qfloat16 value is finite.

Definition at line 93 of file qfloat16.h.

References b16.

◆ isInf()

bool qfloat16::isInf ( ) const

inlinenoexcept

Since: 5.14

Tests whether this qfloat16 value is an infinity.

Definition at line 91 of file qfloat16.h.

References b16.

◆ isNaN()

bool qfloat16::isNaN ( ) const

inlinenoexcept

Since: 5.14

Tests whether this qfloat16 value is "not a number".

Definition at line 92 of file qfloat16.h.

References b16.

◆ isNormal()

bool qfloat16::isNormal ( ) const

inlineconstexprnoexcept

Since: 5.14

Returns true if this qfloat16 value is finite and in normal form.

See also: qFpClassify()

Definition at line 109 of file qfloat16.h.

References b16.

◆ operator float()

QT_WARNING_POP qfloat16::operator float ( ) const

inlinenoexcept

Definition at line 358 of file qfloat16.h.

Friends And Related Symbol Documentation

◆ operator!=

bool operator!=	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 190 of file qfloat16.h.

◆ operator*

qfloat16 operator*	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 144 of file qfloat16.h.

◆ operator+

qfloat16 operator+	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 142 of file qfloat16.h.

◆ operator- [1/2]

qfloat16 operator- ( qfloat16 a )

friend

Definition at line 135 of file qfloat16.h.

◆ operator- [2/2]

qfloat16 operator-	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 143 of file qfloat16.h.

◆ operator/

qfloat16 operator/	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 145 of file qfloat16.h.

◆ operator<

bool operator<	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 186 of file qfloat16.h.

◆ operator<< [1/2]

QT_WARNING_POP friend Q_CORE_EXPORT QDataStream & operator<<	(	QDataStream &	ds,
		qfloat16	f
	)

friend

Definition at line 378 of file qfloat16.cpp.

◆ operator<< [2/2]

Q_CORE_EXPORT QTextStream & operator<<	(	QTextStream &	ts,
		qfloat16	f
	)

friend

Definition at line 409 of file qfloat16.cpp.

◆ operator<=

bool operator<=	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 188 of file qfloat16.h.

◆ operator==

bool operator==	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 189 of file qfloat16.h.

◆ operator>

QT_WARNING_PUSH QT_WARNING_DISABLE_FLOAT_COMPARE friend bool operator>	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 185 of file qfloat16.h.

◆ operator>=

bool operator>=	(	qfloat16	a,
		qfloat16	b
	)

friend

Definition at line 187 of file qfloat16.h.

◆ operator>> [1/2]

Q_CORE_EXPORT QDataStream & operator>>	(	QDataStream &	ds,
		qfloat16 &	f
	)

friend

Definition at line 395 of file qfloat16.cpp.

◆ operator>> [2/2]

Q_CORE_EXPORT QTextStream & operator>>	(	QTextStream &	ts,
		qfloat16 &	f
	)

friend

Definition at line 401 of file qfloat16.cpp.

◆ qFloatFromFloat16()

Q_CORE_EXPORT void qFloatFromFloat16	(	float *	out,
		const qfloat16 *	in,
		qsizetype	len
	)

Since: 5.11

Converts len qfloat16 from in to floats and stores them in out. Both in and out must have len allocated entries.

This function is faster than converting values one by one, and will do runtime F16C detection on x86 and x86-64 hardware.

Definition at line 357 of file qfloat16.cpp.

References hasFastF16(), i, out, and qFloatFromFloat16_fast().

Here is the call graph for this function:

◆ qFloatToFloat16()

Q_CORE_EXPORT void qFloatToFloat16	(	qfloat16 *	out,
		const float *	in,
		qsizetype	len
	)

Since: 5.11

Converts len floats from in to qfloat16 and stores them in out. Both in and out must have len allocated entries.

This function is faster than converting values one by one, and will do runtime F16C detection on x86 and x86-64 hardware.

Definition at line 338 of file qfloat16.cpp.

References hasFastF16(), i, out, and qFloatToFloat16_fast().

Here is the call graph for this function:

◆ qFpClassifyqFpClassify()

int qFpClassifyqFpClassify ( qfloat16 val )

Since: 5.14

Returns the floating-point class of val.

◆ qFuzzyCompareqFuzzyCompare()

bool qFuzzyCompareqFuzzyCompare	(	qfloat16	p1,
		qfloat16	p2
	)

The two numbers are compared in a relative way, where the exactness is stronger the smaller the numbers are.

◆ qIsFiniteqIsFinite()

bool qIsFiniteqIsFinite ( qfloat16 f )

◆ qIsInfqIsInf()

bool qIsInfqIsInf ( qfloat16 f )

◆ qIsNaNqIsNaN()

bool qIsNaNqIsNaN ( qfloat16 f )

◆ qIsNull

bool qIsNull ( qfloat16 f )

friend

Definition at line 308 of file qfloat16.h.

◆ qRound64qRound64()

qint64 qRound64qRound64 ( qfloat16 value )

◆ qRoundqRound()

int qRoundqRound ( qfloat16 value )

Member Data Documentation

◆ b16

quint16 qfloat16::b16

Definition at line 119 of file qfloat16.h.

Referenced by copySign(), isFinite(), isInf(), isNaN(), and isNormal().

◆ IsNative

constexpr bool qfloat16::IsNative = QFLOAT16_IS_NATIVE

staticconstexpr

Definition at line 74 of file qfloat16.h.

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

qtbase/src/corelib/global/qfloat16.h
qtbase/src/corelib/global/qfloat16.cpp
qtbase/src/corelib/global/qfloat16tables.cpp

Public Types

Public Member Functions

Static Public Member Functions

Static Public Attributes

Friends

Related Symbols

Detailed Description

Member Typedef Documentation

◆ NativeType

◆ NearestFloat

Constructor & Destructor Documentation

◆ qfloat16() [1/4]

◆ qfloat16() [2/4]

◆ qfloat16() [3/4]

◆ qfloat16() [4/4]

Member Function Documentation

◆ _limit_denorm_min()

◆ _limit_epsilon()

◆ _limit_infinity()

◆ _limit_lowest()

◆ _limit_max()

◆ _limit_min()

◆ _limit_quiet_NaN()

◆ copySign()

◆ fpClassify()

◆ isFinite()

◆ isInf()

◆ isNaN()

◆ isNormal()

◆ operator float()

Friends And Related Symbol Documentation

◆ operator!=

◆ operator*

◆ operator+

◆ operator- [1/2]

◆ operator- [2/2]

◆ operator/

◆ operator<

◆ operator<< [1/2]

◆ operator<< [2/2]

◆ operator<=

◆ operator==

◆ operator>

◆ operator>=

◆ operator>> [1/2]

◆ operator>> [2/2]

◆ qFloatFromFloat16()

◆ qFloatToFloat16()

◆ qFpClassifyqFpClassify()

◆ qFuzzyCompareqFuzzyCompare()

◆ qIsFiniteqIsFinite()

◆ qIsInfqIsInf()

◆ qIsNaNqIsNaN()

◆ qIsNull

◆ qRound64qRound64()

◆ qRoundqRound()

Member Data Documentation

◆ b16

◆ IsNative