Handling Floating-Point Exceptions in FORTRAN

ID: Q57918

The information in this article applies to:

Microsoft FORTRAN for MS-DOS, versions 3.3, 3.31, 4.0, 4.01, 4.1, 5.0, 5.1
Microsoft FORTRAN for OS/2, versions 4.1, 5.0, 5.1
Microsoft FORTRAN PowerStation for MS-DOS, versions 1.0, 1.0a
Microsoft Fortran Powerstation 32 for Windows NT, versions 1.0, 4.0

SUMMARY

The IEEE floating point standard defines the following five exceptions: invalid operation, divide by zero, overflow, underflow, and precision. The Microsoft FORTRAN run-time libraries provide the following three routines to obtain and modify the value of the status and control words:

The store-status-word (SSWRQQ) function returns the value of the status word.

The store-control-word (SCWRQQ) function returns the value of the control word.

The load-control-word (LCWRQQ) function sets the control word to a specified value.

An application can specify the manner to handle an exception by changing the control word and can determine which exceptions have occurred by checking the value of the status word.

MORE INFORMATION

The following code example demonstrates using the LCWRQQ function to disable the overflow and divide by zero exceptions and their corresponding run-time errors.

Sample Code


C Compile options needed: None

      INTERFACE TO SUBROUTINE LCWRQQ(CW)
      INTEGER*2 CW
      END

      PROGRAM NOEXCEPTIONS

      INTEGER*2 CW
      PARAMETER(CW = 16#133F)
      REAL*4 X, Y, Z

      X = 1.0
      Y = 0.0

      CALL LCWRQQ(CW)

      Z = X / Y

      END

NOTE: In FORTRAN PowerStation, GETSTATUSFPQQ is equivalent to SSWRQQ, GETCONTROLFPQQ is equivalent to SCWRQQ, and SETCONTROLFPQQ is equivalent to LCWRQQ.

Additional query words: kbinf 1.00 3.30 4.00 4.10 5.00 5.10

Keywords : kbcode kbFortranPS kbLangFortran
Version : :1.0,1.0a,3.3,3.31,4.0,4.01,4.1,5.0,5.1
Platform : MS-DOS NT OS/2 WINDOWS
Issue type :