Class for analyzing a floating point number. More...

Public Member Functions
	FPInfo (void) noexcept
	Default constructor.

	FPInfo (uint16_t uValue) noexcept
	Constructor for 16 bit float.

	FPInfo (float fValue) noexcept
	Constructor for 32 bit float.

	FPInfo (double dValue) noexcept
	Constructor for 64 bit float.

void	init_half (uint16_t uValue) noexcept
	Process a 16 bit floating point number.

void	init_float (float fValue) noexcept
	Process a 32 bit floating point number.

void	init_double (double dValue) noexcept
	Process a 64 bit floating point number.

uint64_t	get_mantissa (void) const noexcept
	Get the mantissa value.

uint32_t	get_biased_exponent (void) const noexcept
	Get the biased exponent.

int32_t	get_exponent (void) const noexcept
	Get the exponent.

uint32_t	get_mantissa_bit_count (void) const noexcept
	Get the number of bits in the mantissa.

uint32_t	get_exponent_bit_count (void) const noexcept
	Get the number of bits in the exponent.

uint_t	is_valid (void) const noexcept
	Detect if a number was analyzed.

uint_t	is_negative (void) const noexcept
	Detect if a number is a negative value.

uint_t	is_zero (void) const noexcept
	Detect if a number is zero.

uint_t	is_NaN (void) const noexcept
	Detect if a number is Not a Number.

uint_t	is_QNaN (void) const noexcept
	Detect if a number is a quiet Not a Number.

uint_t	is_indefinite_NaN (void) const noexcept
	Detect if a number is an indefinite Not a Number.

uint_t	is_infinity (void) const noexcept
	Detect if a number is infinity.

uint_t	is_subnormal (void) const noexcept
	Detect if a number is a non normalized number.

uint_t	is_finite (void) const noexcept
	Detect if a number is a finite number.

FPInfo &	operator= (uint16_t usValue) noexcept
	Process a 16 bit floating point number.

FPInfo &	operator= (float fValue) noexcept
	Process a 32 bit floating point number.

FPInfo &	operator= (double dValue) noexcept
	Process a 64 bit floating point number.

Private Attributes
uint64_t	m_uMantissa
	The mantissa value for the floating point number.

uint32_t	m_uBiasedExponent
	Exponent as stored in the floating point number.

int32_t	m_iExponent
	Exponent with bias applied.

uint32_t	m_uMantissaBitCount
	Number of bits in the mantissa.

uint32_t	m_uExponentBitCount
	Number of bits in the exponent.

uint32_t	m_uFlags
	Flags that contain the determined states.

Static Private Attributes
static constexpr const uint32_t	kSignBit16 = 0x8000U
	Location of the sign bit in a 16 bit float.

static constexpr const uint32_t	kExponentMask16 = 0x7C00U
	Mask for the exponent in a 16 bit float.

static constexpr const uint32_t	kMantissaMask16 = 0x03FFU
	Mask for the mantissa in a 16 bit float.

static constexpr const uint32_t	kImplicitBit16 = 0x0400U
	Implicit "1" bit for a 16 bit float.

static constexpr const uint32_t	kMantissaBitCount16 = 10
	Number of bits in the mantissa for a 16 bit float.

static constexpr const uint32_t	kExponentBitCount16 = 5
	Number of bits in the exponent for a 16 bit float.

static constexpr const uint32_t	kExponentBias16 = 15
	Bias of the exponent for a 16 bit float.

static constexpr const uint32_t	kSpecialExponent16 = 0x1FU
	Special exponent for INF/NAN in a 16 bit float.

static constexpr const uint32_t	kNANTypeBit16 = 0x0200U
	QNAN or SNAN bit in a 16 bit float.

static constexpr const uint32_t	kSignBit32 = 0x80000000U
	Location of the sign bit in a 32 bit float.

static constexpr const uint32_t	kExponentMask32 = 0x7F800000U
	Mask for the exponent in a 32 bit float.

static constexpr const uint32_t	kMantissaMask32 = 0x007FFFFFU
	Mask for the mantissa in a 32 bit float.

static constexpr const uint32_t	kImplicitBit32 = 0x00800000U
	Implicit "1" bit for a 32 bit float.

static constexpr const uint32_t	kMantissaBitCount32 = 23
	Number of bits in the mantissa for a 32 bit float.

static constexpr const uint32_t	kExponentBitCount32 = 8
	Number of bits in the exponent for a 32 bit float.

static constexpr const uint32_t	kExponentBias32 = 127
	Bias of the exponent for a 32 bit float.

static constexpr const uint32_t	kSpecialExponent32 = 0xFFU
	Special exponent for INF/NAN in a 32 bit float.

static constexpr const uint32_t	kNANTypeBit32 = 0x00400000U
	QNAN or SNAN bit in a 32 bit float.

static constexpr const uint64_t	kSignBit64 = 0x8000000000000000ULL
	Location of the sign bit in a 64 bit float.

static constexpr const uint64_t	kExponentMask64
	Mask for the exponent in a 64 bit float.

static constexpr const uint64_t	kMantissaMask64
	Mask for the mantissa in a 64 bit float.

static constexpr const uint64_t	kImplicitBit64
	Implicit "1" bit for a 64 bit float.

static constexpr const uint32_t	kMantissaBitCount64 = 52
	Number of bits in the mantissa for a 64 bit float.

static constexpr const uint32_t	kExponentBitCount64 = 11
	Number of bits in the exponent for a 64 bit float.

static constexpr const uint32_t	kExponentBias64 = 1023
	Bias of the exponent for a 64 bit float.

static constexpr const uint32_t	kSpecialExponent64 = 0x7FFU
	Special exponent for INF/NAN in a 64 bit float.

static constexpr const uint64_t	kNANTypeBit64
	QNAN or SNAN bit in a 64 bit float.

static constexpr const uint32_t	kFlagValidShift = 0U
	Flag that's set if a floating point number was analyzed.

static constexpr const uint32_t	kFlagNegativeShift = 1U
	Flag that's set if the number is negative.

static constexpr const uint32_t	kFlagZeroShift = 2U
	Flag that's set if the number is zero.

static constexpr const uint32_t	kFlagNANShift = 3U
	Flag that's set if the number is a "Not a Number".

static constexpr const uint32_t	kFlagQNANShift = 4U
	Flag that's set if the number is a quiet "Not a Number".

static constexpr const uint32_t	kFlagIndefiniteNANShift = 5U
	Flag that's set if the number is an unknown "Not a Number".

static constexpr const uint32_t	kFlagInfinityShift = 6U
	Flag that's set if the number is infinite.

static constexpr const uint32_t	kFlagSubNormalShift = 7U
	Flag that's set if the number is not normalized.

Detailed Description

Class for analyzing a floating point number.

Submitting a number into this class causes it to be broken down into its components so it can be queried as to the floating point behaviors this number represents.

It's useful for printing functions or number analysis functions.

For lightweight tests, use the explicit calls such at is_NaN(float), is_infinite(double), etc

When a floating point number is assigned to this class, it is broken down to the type of value (Infinity, NaN, Subnormal), the mantissa, the exponent and the sign. The values can be accessed with the calls get_mantissa(), get_exponent(), get_mantissa_bit_count(), etc.

See also: is_NaN(float), is_NaN(), get_mantissa(), get_exponent()

Constructor & Destructor Documentation

◆ FPInfo() [1/4]

Burger::FPInfo::FPInfo ( void )

explicitnoexcept

Default constructor.

All values are set to zero. Use the Init() functions or assignment operators to perform floating point analysis.

See also: init_float(float), or FPInfo::operator = (float)

◆ FPInfo() [2/4]

Burger::FPInfo::FPInfo ( uint16_t uValue )

explicitnoexcept

Constructor for 16 bit float.

Parameters

uValue 16 bit floating point number

See also: init_half(uint16_t), FPInfo(double) or FPInfo(float)

◆ FPInfo() [3/4]

Burger::FPInfo::FPInfo ( float fValue )

explicitnoexcept

Constructor for 32 bit float.

Parameters

fValue 32 bit floating point number

See also: init_float(float), FPInfo(uint16_t) or FPInfo(double)

◆ FPInfo() [4/4]

Burger::FPInfo::FPInfo ( double dValue )

explicitnoexcept

Constructor for 64 bit float.

Parameters

dValue 64 bit floating point number

See also: init_double(double), FPInfo(uint16_t) or FPInfo(float)

Member Function Documentation

◆ get_biased_exponent()

uint32_t Burger::FPInfo::get_biased_exponent ( void ) const

inlinenoexcept

Get the biased exponent.

Returns: The exponent value as it was stored in the floating point number

◆ get_exponent()

int32_t Burger::FPInfo::get_exponent ( void ) const

inlinenoexcept

Get the exponent.

Returns the exponent as it was meant to be represented, as such, it's a signed integer.

Returns: The exponent value of the floating point number

◆ get_exponent_bit_count()

uint32_t Burger::FPInfo::get_exponent_bit_count ( void ) const

inlinenoexcept

Get the number of bits in the exponent.

Returns: The exponent size in bits of the analyzed floating point number

◆ get_mantissa()

uint64_t Burger::FPInfo::get_mantissa ( void ) const

inlinenoexcept

Get the mantissa value.

Returns: The mantissa of the analyzed floating point number

◆ get_mantissa_bit_count()

uint32_t Burger::FPInfo::get_mantissa_bit_count ( void ) const

inlinenoexcept

Get the number of bits in the mantissa.

Returns: The mantissa size in bits of the analyzed floating point number

◆ init_double()

void BURGER_API Burger::FPInfo::init_double ( double dValue )

noexcept

Process a 64 bit floating point number.

Given a 64 bit floating point number, process it into its component parts.

Parameters

dValue 64 bit float point number

◆ init_float()

void BURGER_API Burger::FPInfo::init_float ( float fValue )

noexcept

Process a 32 bit floating point number.

Given a 32 bit floating point number, process it into its component parts.

Parameters

fValue 32 bit float point number

◆ init_half()

void BURGER_API Burger::FPInfo::init_half ( uint16_t uValue )

noexcept

Process a 16 bit floating point number.

Given a 16 bit floating point number, process it into its component parts.

Parameters

uValue 16 bit float point number

◆ is_finite()

uint_t Burger::FPInfo::is_finite ( void ) const

inlinenoexcept

Detect if a number is a finite number.

If the number is infinity or NaN, it will return FALSE. Otherwise it will return TRUE.

Returns: TRUE if the number is finite, FALSE if it is.

See also: is_finite(float) or is_finite(double)

◆ is_indefinite_NaN()

uint_t Burger::FPInfo::is_indefinite_NaN ( void ) const

inlinenoexcept

Detect if a number is an indefinite Not a Number.

Returns: TRUE if the number is an indefinite Not a Number, FALSE if it's a valid number

◆ is_infinity()

uint_t Burger::FPInfo::is_infinity ( void ) const

inlinenoexcept

Detect if a number is infinity.

Returns: TRUE if the number is infinity, FALSE if it's not

See also: is_infinite(float) or is_infinite(double)

◆ is_NaN()

uint_t Burger::FPInfo::is_NaN ( void ) const

inlinenoexcept

Detect if a number is Not a Number.

This returns TRUE for all forms of Not a Number.

Returns: TRUE if the number is Not a Number, FALSE if it's a valid number

See also: is_NaN(float) or is_NaN(double)

◆ is_negative()

uint_t Burger::FPInfo::is_negative ( void ) const

inlinenoexcept

Detect if a number is a negative value.

Returns: TRUE if the number is negative, FALSE if not

◆ is_QNaN()

uint_t Burger::FPInfo::is_QNaN ( void ) const

inlinenoexcept

Detect if a number is a quiet Not a Number.

Returns: TRUE if the number is a quiet Not a Number, FALSE if it's a valid number

◆ is_subnormal()

uint_t Burger::FPInfo::is_subnormal ( void ) const

inlinenoexcept

Detect if a number is a non normalized number.

Returns: TRUE if the number is not normalized, FALSE if it is.

◆ is_valid()

uint_t Burger::FPInfo::is_valid ( void ) const

inlinenoexcept

Detect if a number was analyzed.

Returns: TRUE if an analyzed number is stored in the class, FALSE if not.

◆ is_zero()

uint_t Burger::FPInfo::is_zero ( void ) const

inlinenoexcept

Detect if a number is zero.

This returns TRUE for both positive and negative zero.

Returns: TRUE if the number is zero, FALSE if not

◆ operator=() [1/3]

FPInfo & Burger::FPInfo::operator= ( double dValue )

inlinenoexcept

Process a 64 bit floating point number.

Given a 64 bit floating point number, process it into its component parts.

Parameters

dValue 64 bit float point number

◆ operator=() [2/3]

FPInfo & Burger::FPInfo::operator= ( float fValue )

inlinenoexcept

Process a 32 bit floating point number.

Given a 32 bit floating point number, process it into its component parts.

Parameters

fValue 32 bit float point number

◆ operator=() [3/3]

FPInfo & Burger::FPInfo::operator= ( uint16_t usValue )

inlinenoexcept

Process a 16 bit floating point number.

Given a 16 bit floating point number, process it into its component parts.

Parameters

usValue 16 bit float point number

Member Data Documentation

◆ kExponentBias16

const uint32_t Burger::FPInfo::kExponentBias16 = 15

staticconstexprprivate

Bias of the exponent for a 16 bit float.

◆ kExponentBias32

const uint32_t Burger::FPInfo::kExponentBias32 = 127

staticconstexprprivate

Bias of the exponent for a 32 bit float.

◆ kExponentBias64

const uint32_t Burger::FPInfo::kExponentBias64 = 1023

staticconstexprprivate

Bias of the exponent for a 64 bit float.

◆ kExponentBitCount16

const uint32_t Burger::FPInfo::kExponentBitCount16 = 5

staticconstexprprivate

Number of bits in the exponent for a 16 bit float.

◆ kExponentBitCount32

const uint32_t Burger::FPInfo::kExponentBitCount32 = 8

staticconstexprprivate

Number of bits in the exponent for a 32 bit float.

◆ kExponentBitCount64

const uint32_t Burger::FPInfo::kExponentBitCount64 = 11

staticconstexprprivate

Number of bits in the exponent for a 64 bit float.

◆ kExponentMask16

const uint32_t Burger::FPInfo::kExponentMask16 = 0x7C00U

staticconstexprprivate

Mask for the exponent in a 16 bit float.

◆ kExponentMask32

const uint32_t Burger::FPInfo::kExponentMask32 = 0x7F800000U

staticconstexprprivate

Mask for the exponent in a 32 bit float.

◆ kExponentMask64

const uint64_t Burger::FPInfo::kExponentMask64

staticconstexprprivate

Initial value:

=

0x7FF0000000000000ULL

Mask for the exponent in a 64 bit float.

◆ kFlagIndefiniteNANShift

const uint32_t Burger::FPInfo::kFlagIndefiniteNANShift = 5U

staticconstexprprivate

Flag that's set if the number is an unknown "Not a Number".

◆ kFlagInfinityShift

const uint32_t Burger::FPInfo::kFlagInfinityShift = 6U

staticconstexprprivate

Flag that's set if the number is infinite.

◆ kFlagNANShift

const uint32_t Burger::FPInfo::kFlagNANShift = 3U

staticconstexprprivate

Flag that's set if the number is a "Not a Number".

◆ kFlagNegativeShift

const uint32_t Burger::FPInfo::kFlagNegativeShift = 1U

staticconstexprprivate

Flag that's set if the number is negative.

◆ kFlagQNANShift

const uint32_t Burger::FPInfo::kFlagQNANShift = 4U

staticconstexprprivate

Flag that's set if the number is a quiet "Not a Number".

◆ kFlagSubNormalShift

const uint32_t Burger::FPInfo::kFlagSubNormalShift = 7U

staticconstexprprivate

Flag that's set if the number is not normalized.

◆ kFlagValidShift

const uint32_t Burger::FPInfo::kFlagValidShift = 0U

staticconstexprprivate

Flag that's set if a floating point number was analyzed.

◆ kFlagZeroShift

const uint32_t Burger::FPInfo::kFlagZeroShift = 2U

staticconstexprprivate

Flag that's set if the number is zero.

◆ kImplicitBit16

const uint32_t Burger::FPInfo::kImplicitBit16 = 0x0400U

staticconstexprprivate

Implicit "1" bit for a 16 bit float.

◆ kImplicitBit32

const uint32_t Burger::FPInfo::kImplicitBit32 = 0x00800000U

staticconstexprprivate

Implicit "1" bit for a 32 bit float.

◆ kImplicitBit64

const uint64_t Burger::FPInfo::kImplicitBit64

staticconstexprprivate

Initial value:

=

0x0010000000000000ULL

Implicit "1" bit for a 64 bit float.

◆ kMantissaBitCount16

const uint32_t Burger::FPInfo::kMantissaBitCount16 = 10

staticconstexprprivate

Number of bits in the mantissa for a 16 bit float.

◆ kMantissaBitCount32

const uint32_t Burger::FPInfo::kMantissaBitCount32 = 23

staticconstexprprivate

Number of bits in the mantissa for a 32 bit float.

◆ kMantissaBitCount64

const uint32_t Burger::FPInfo::kMantissaBitCount64 = 52

staticconstexprprivate

Number of bits in the mantissa for a 64 bit float.

◆ kMantissaMask16

const uint32_t Burger::FPInfo::kMantissaMask16 = 0x03FFU

staticconstexprprivate

Mask for the mantissa in a 16 bit float.

◆ kMantissaMask32

const uint32_t Burger::FPInfo::kMantissaMask32 = 0x007FFFFFU

staticconstexprprivate

Mask for the mantissa in a 32 bit float.

◆ kMantissaMask64

const uint64_t Burger::FPInfo::kMantissaMask64

staticconstexprprivate

Initial value:

=

0x000FFFFFFFFFFFFFULL

Mask for the mantissa in a 64 bit float.

◆ kNANTypeBit16

const uint32_t Burger::FPInfo::kNANTypeBit16 = 0x0200U

staticconstexprprivate

QNAN or SNAN bit in a 16 bit float.

◆ kNANTypeBit32

const uint32_t Burger::FPInfo::kNANTypeBit32 = 0x00400000U

staticconstexprprivate

QNAN or SNAN bit in a 32 bit float.

◆ kNANTypeBit64

const uint64_t Burger::FPInfo::kNANTypeBit64

staticconstexprprivate

Initial value:

=

0x0008000000000000ULL

QNAN or SNAN bit in a 64 bit float.

◆ kSignBit16

const uint32_t Burger::FPInfo::kSignBit16 = 0x8000U

staticconstexprprivate

Location of the sign bit in a 16 bit float.

◆ kSignBit32

const uint32_t Burger::FPInfo::kSignBit32 = 0x80000000U

staticconstexprprivate

Location of the sign bit in a 32 bit float.

◆ kSignBit64

const uint64_t Burger::FPInfo::kSignBit64 = 0x8000000000000000ULL

staticconstexprprivate

Location of the sign bit in a 64 bit float.

◆ kSpecialExponent16

const uint32_t Burger::FPInfo::kSpecialExponent16 = 0x1FU

staticconstexprprivate

Special exponent for INF/NAN in a 16 bit float.

◆ kSpecialExponent32

const uint32_t Burger::FPInfo::kSpecialExponent32 = 0xFFU

staticconstexprprivate

Special exponent for INF/NAN in a 32 bit float.

◆ kSpecialExponent64

const uint32_t Burger::FPInfo::kSpecialExponent64 = 0x7FFU

staticconstexprprivate

Special exponent for INF/NAN in a 64 bit float.

◆ m_iExponent

int32_t Burger::FPInfo::m_iExponent

private

Exponent with bias applied.

◆ m_uBiasedExponent

uint32_t Burger::FPInfo::m_uBiasedExponent

private

Exponent as stored in the floating point number.

◆ m_uExponentBitCount

uint32_t Burger::FPInfo::m_uExponentBitCount

private

Number of bits in the exponent.

◆ m_uFlags

uint32_t Burger::FPInfo::m_uFlags

private

Flags that contain the determined states.

◆ m_uMantissa

uint64_t Burger::FPInfo::m_uMantissa

private

The mantissa value for the floating point number.

◆ m_uMantissaBitCount

uint32_t Burger::FPInfo::m_uMantissaBitCount

private

Number of bits in the mantissa.

Public Member Functions

Private Attributes

Static Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ FPInfo() [1/4]

◆ FPInfo() [2/4]

◆ FPInfo() [3/4]

◆ FPInfo() [4/4]

Member Function Documentation

◆ get_biased_exponent()

◆ get_exponent()

◆ get_exponent_bit_count()

◆ get_mantissa()

◆ get_mantissa_bit_count()

◆ init_double()

◆ init_float()

◆ init_half()

◆ is_finite()

◆ is_indefinite_NaN()

◆ is_infinity()

◆ is_NaN()

◆ is_negative()

◆ is_QNaN()

◆ is_subnormal()

◆ is_valid()

◆ is_zero()

◆ operator=() [1/3]

◆ operator=() [2/3]

◆ operator=() [3/3]

Member Data Documentation

◆ kExponentBias16

◆ kExponentBias32

◆ kExponentBias64

◆ kExponentBitCount16

◆ kExponentBitCount32

◆ kExponentBitCount64

◆ kExponentMask16

◆ kExponentMask32

◆ kExponentMask64

◆ kFlagIndefiniteNANShift

◆ kFlagInfinityShift

◆ kFlagNANShift

◆ kFlagNegativeShift

◆ kFlagQNANShift

◆ kFlagSubNormalShift

◆ kFlagValidShift

◆ kFlagZeroShift

◆ kImplicitBit16

◆ kImplicitBit32

◆ kImplicitBit64

◆ kMantissaBitCount16

◆ kMantissaBitCount32

◆ kMantissaBitCount64

◆ kMantissaMask16

◆ kMantissaMask32

◆ kMantissaMask64

◆ kNANTypeBit16

◆ kNANTypeBit32

◆ kNANTypeBit64

◆ kSignBit16

◆ kSignBit32

◆ kSignBit64

◆ kSpecialExponent16

◆ kSpecialExponent32

◆ kSpecialExponent64

◆ m_iExponent

◆ m_uBiasedExponent

◆ m_uExponentBitCount

◆ m_uFlags

◆ m_uMantissa

◆ m_uMantissaBitCount