frysk.value
Class FloatingPoint854Format

java.lang.Object
  frysk.value.FloatingPointFormat
      frysk.value.FloatingPoint854Format

public class FloatingPoint854Format
extends FloatingPointFormat

Field Summary

static FloatingPointFormat	IEEE128
static FloatingPointFormat	IEEE32
static FloatingPointFormat	IEEE64
static FloatingPointFormat	IEEE80
static FloatingPointFormat	IEEE96
private int	integralOfMantissa
private Packing	packExponent
private Packing	packFraction
private int	size
private int	sizeE
private int	sizeF
private static BigDecimal	two

Constructor Summary

FloatingPoint854Format(int size, int idxE, int sizeE, int idxF, int sizeF)

Method Summary

private BigDecimal	divide(BigDecimal a, BigDecimal b)
(package private) BigInteger	getBiasedExponent(byte[] bytes)
(package private) BigInteger	getFraction(byte[] bytes)
(package private) int	getIntegralOfMantissa(byte[] bytes)
private BigDecimal	getMantissa(BigInteger f, BigInteger e, int sizeOfF) Gets mantissa value according to IEEE 754/854 floating point rules
private BigDecimal	getMantissaExtended(BigInteger f, int sizeOfF)
(package private) BigInteger	getMaxEValue()
(package private) int	getSign(byte[] bytes)
(package private) byte[]	pack(BigFloatingPoint value, int size) FIXME: Convert to byte[] using Packing methods.
private BigFloatingPoint	toBigFP(int s, BigInteger e, BigInteger f, BigInteger maxE) Calculate floating point value according to IEEE 754/854 rules.
(package private) BigFloatingPoint	unpack(byte[] bytes)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

IEEE32

public static final FloatingPointFormat IEEE32

IEEE64

public static final FloatingPointFormat IEEE64

IEEE128

public static final FloatingPointFormat IEEE128

IEEE80

public static final FloatingPointFormat IEEE80

IEEE96

public static final FloatingPointFormat IEEE96

packExponent

private Packing packExponent

packFraction

private Packing packFraction

sizeF

private int sizeF

sizeE

private int sizeE

size

private int size

integralOfMantissa

private int integralOfMantissa

two

private static final BigDecimal two

Constructor Detail

FloatingPoint854Format

FloatingPoint854Format(int size,
                       int idxE,
                       int sizeE,
                       int idxF,
                       int sizeF)

Parameters:

size - - size of floating point in bytes

idxE - - begin bit index of exponent field

sizeE - - bit size of exponent field

idxF - - begin bit index of fraction field

sizeF - - size of fraction field Note: include j bit in fraction field, where applicable.

Method Detail

unpack

BigFloatingPoint unpack(byte[] bytes)

Specified by:

unpack in class FloatingPointFormat

getSign

int getSign(byte[] bytes)

Returns:

0 for negative and 1 for positive values FPs.

getBiasedExponent

BigInteger getBiasedExponent(byte[] bytes)

getFraction

BigInteger getFraction(byte[] bytes)

getIntegralOfMantissa

int getIntegralOfMantissa(byte[] bytes)

getMaxEValue

BigInteger getMaxEValue()

getMantissa

private BigDecimal getMantissa(BigInteger f,
                               BigInteger e,
                               int sizeOfF)

Gets mantissa value according to IEEE 754/854 floating point rules

Parameters:

f - - value of fraction field (including j bit where applicable)

e - - value of exponent field

sizeOfF - - bit size of fraction field (including j bit where applicable)

Returns:

mantissa

getMantissaExtended

private BigDecimal getMantissaExtended(BigInteger f,
                                       int sizeOfF)

toBigFP

private BigFloatingPoint toBigFP(int s,
                                 BigInteger e,
                                 BigInteger f,
                                 BigInteger maxE)

Calculate floating point value according to IEEE 754/854 rules.

Parameters:

s - - sign bit

e - - value of exponent field

m - - mantissa

maxE - - max possible value of exponent field

divide

private BigDecimal divide(BigDecimal a,
                          BigDecimal b)

pack

byte[] pack(BigFloatingPoint value,
            int size)

FIXME: Convert to byte[] using Packing methods. Need to store mantissa and exponent value?

Specified by:

pack in class FloatingPointFormat