FIX: FP Emulator Converts REAL to INTEGER Incorrectly

MORE INFORMATION

The FORTRAN PowerStation floating-point emulator is a virtual dynamic- link library (DLL) in the Phar Lap MS-DOS extender. To duplicate this problem on a computer with a math coprocessor installed, type the following at the MS-DOS command prompt:

This statement causes the MS-DOS extender to ignore the math coprocessor and use the emulator code for all math operations.

The following code example prints the word "Failed" when it runs with the floating-point emulator:

Sample Code 1

C Compiler options needed: none

      x =  3.0
      z = -3. ** x
      i =  z
      if (i .ne. -27) then
        print *, 'Failed'
      else
        print *, 'OK'
      end if
      end
				

The following code example demonstrates using the NINT intrinsic function to avoid this error. This code prints the word "OK" even when a math coprocessor is not available.

Sample Code 2

C Compiler options needed: none

      x =  3.0
      z = -3. ** x
      i =  nint(z)
      if (i .ne. -27) then
        print *, 'Failed'
      else
        print *, 'OK'
      end if
      end