intro_ieee

Date:: 10-12-2011

NAME

INTRO_IEEE - Introduction to Fortran standard modules and procedures for IEEE

DESCRIPTION

Three intrinsic Fortran modules provide support for IEEE exceptions and arithmetic: IEEE_EXCEPTIONS, IEEE_ARITHMETIC, and IEEE_FEATURES. These modules provide IEEE procedures and named constants that can be used to check for IEEE exceptions and arithmetic support.

IEEE_EXCEPTIONS Module

The IEEE_EXCEPTIONS module defines the following named constants for the exceptions:

IEEE_DIVIDE_BY_ZERO
IEEE_INEXACT
IEEE_INVALID
IEEE_OVERFLOW
IEEE_UNDERFLOW

The IEEE_EXCEPTIONS module defines an IEEE_FLAG_TYPE and an IEEE_STATUS_TYPE.

The IEEE_FLAG_TYPE is provided for identifying an exception flag; its value is one of the named constants for the IEEE exceptions. The array named constant IEEE_USUAL contains elements IEEE_OVERFLOW, IEEE_DIVIDE_BY_ZERO, and IEEE_INVALID.

The array named contant IEEE_ALL contains elements IEEE_USUAL, IEEE_UNDERFLOW, and IEEE_INEXACT.

The IEEE_STATUS_TYPE is provided for saving the current floating-point status.

IEEE_ARITHMETIC Module

The IEEE_ARITHMETIC module defines IEEE_CLASS_TYPE to identify a class of floating point values. The named constants for IEEE_CLASS_TYPE are:

IEEE_SIGNALING_NAN
IEEE_QUIET_NAN
IEEE_NEGATIVE_DENORMAL
IEEE_NEGATIVE_INF
IEEE_NEGATIVE_NORMAL
IEEE_NEGATIVE_ZERO
IEEE_POSITIVE_DENORMAL
IEEE_POSITIVE_INF
IEEE_POSITIVE_NORMAL
IEEE_POSITIVE_ZERO
IEEE_OTHER_VALUE

The IEEE_ARITHMETIC module defines IEEE_ROUND_TYPE to identify a particular IEEE rounding mode. The named constants for IEEE_ROUNDS_TYPE are:

IEEE_DOWN
IEEE_NEAREST
IEEE_OTHER
IEEE_TO_ZERO
IEEE_UP

The IEEE_ARITHMETIC module defines the == operator to determine if two values of type IEEE_ROUND_TYPE or IEEE_CLASS_TYPE are the same. The operation returns true if they are the same and false otherwise.

The IEEE_ARITHMETIC module defines the /= operator to determine if two values of type IEEE_ROUND_TYPE or IEEE_CLASS_TYPE differ. The operation returns true if they differ and false otherwise.

IEEE_FEATURES Module

The IEEE_FEATURES module defines IEEE_FEATURES_TYPE to identify the status of a particular IEEE feature. The named constants for this type are:

IEEE_DATATYPE
IEEE_DENORMAL
IEEE_DIVIDE
IEEE_HALTING
IEEE_INEXACT_FLAG
IEEE_INF
IEEE_INVALID_FLAG
IEEE_NAN
IEEE_ROUNDING
IEEE_SQRT
IEEE_UNDERFLOW_FLAG

The IEEE procedures consist of functions and subroutines.

Inquiry Functions

The IEEE_EXCEPTIONS module contains the following inquiry functions:

IEEE_SUPPORT_FLAG(FLAG[,X]): Are IEEE exceptions supported?
IEEE_SUPPORT_HALTING(FLAG): Is IEEE halting control supported?

The IEEE_ARITHMETIC module contains the following inquiry functions:

IEEE_SUPPORT_DATATYPE([X]): Is IEEE arithmetic supported?
IEEE_SUPPORT_DENORMAL([X]): Are IEEE denormalized numbers supported?
IEEE_SUPPORT_DIVIDE([X]): Is IEEE divide supported?
IEEE_SUPPORT_INF([X]): Is IEEE infinity supported?
IEEE_SUPPORT_IO([X]): Supports IEEE base conversion rounding during formatted input/output?
IEEE_SUPPORT_NAN([X]): Are IEEE NaNs supported?
IEEE_SUPPORT_ROUNDING(ROUND_VALUE[,X]): Is IEEE rounding supported?
IEEE_SUPPORT_SQRT([X]): Is IEEE SQRT supported?
IEEE_SUPPORT_STANDARD([X]): Are all IEEE facilities supported?
IEEE_SUPPORT_UNDERFLOW_CONTROL([X]): Is IEEE underflow control supported?

Elemental Functions

The IEEE_ARITHMETIC module contains the elemental functions for reals X and Y when the IEEE_SUPPORT_DATATYPE(X) and IEEE_SUPPORT_DATATYPE(Y) are true:

IEEE_CLASS(X): Returns IEEE class.
IEEE_COPY_SIGN(X,Y): IEEE copy sign function. Returns the value of X with the sign of Y.
IEEE_IS_FINITE(X): Is the value X finite?
IEEE_IS_NAN(X): Is the value Xa NaN?
IEEE_IS_NORMAL(X): Is the value X normal (that is, neither infinity, NaN, nor denormalized)?
IEEE_IS_NEGATIVE(X): Is the value X negative?
IEEE_LOGB(X): Returns the unbiased exponent of X.
IEEE_NEXT_AFTER(X,Y): Returns the next representable neighbor of X.
IEEE_REM(X,Y): Returns X-Y*N, where N is the integer nearest to the exact value of X/Y.
IEEE_RINT(X): Rounds to integer value according to the current rounding mode.
IEEE_SCALB(X,I): Returns X* 2**I
IEEE_UNORDERED(X,Y): IEEE unordered function. Returns true if X or Y is NaN and false otherwise.
IEEE_VALUE(X,CLASS): Generates an IEEE value.

KIND Function

The IEEE_ARITHMETIC module contains the transformational kind function:

IEEE_SELECTED_REAL_KIND([P,R]): Returns kind type parameter value for an IEEE real with precision P and range R.

Elemental Subroutines

The IEEE_EXCEPTIONS module contains the elemental subroutines:

IEEE_GET_FLAG(FLAG,FLAG_VALUE): Gets an exception flag.
IEEE_GET_HALTING_MODE(FLAG,HALTING): Gets halting mode for an exception.

Nonelemental Subroutines

The IEEE_EXCEPTIONS module contains the nonelemental subroutines:

IEEE_GET_STATUS(STATUS_VALUE): Returns current floating-point status.
IEEE_SET_FLAG(FLAG,FLAG_VALUE): Sets an exception flag.
IEEE_SET_HALTING_MODE(FLAG,HALTING): Controls continuation or halting on exceptions.
IEEE_SET_STATUS(STATUS_VALUE): Restores current state of floating_point environment.

Only IEEE_SET_FLAG and IEEE_SET_HALTING_MODE are pure subroutines in module IEEE_EXCEPTIONS.

The IEEE_ARITHMETIC module contains the nonelemental subroutines:

IEEE_GET_ROUNDING_MODE(ROUND_VALUE): Gets current IEEE rounding mode.
IEEE_GET_UNDERFLOW_MODE(GRADUAL): Gets current underflow mode.
IEEE_SET_ROUNDING_MODE(ROUND_VALUE): Sets current IEEE rounding mode.
IEEE_SET_UNDERFLOW_MODE(GRADUAL): Sets current underflow mode.

The IEEE_ARITHMETIC module does not have any pure nonelemental subroutines.