topic Thanks Ian. the problem is in Intel® Fortran Compiler

strange behaviour from 64-bit integer

Alireza_Forghani — Wed, 14 Jan 2015 20:46:00 GMT

I'm seeing a strange behaviour from int64 integer. Following code:

    use iso_fortran_env
    implicit none

    integer (int64) :: i
    
    i=2**32-2
    
write (*,*) "i, btest (i,0)" , i, btest (i,0)
    write (*,*) "i, btest (i,31)" , i, btest (i,31)
    write (*,*) "i, btest (i,32)" , i, btest (i,32)
    write (*,*) "i, btest (i,33)" , i, btest (i,33)
    write (*,*) "huge (i)" , huge (i)

returns:

i, btest (i,0) -2 F
i, btest (i,31) -2 T
i, btest (i,32) -2 T
i, btest (i,33) -2 T
huge (i) 9223372036854775807

which shows that it calculates 2^32-2 as -2. Whereas limit on int64 is 9223372036854775807 that is way bigger than 2^32!

Wonder if I'm doing something wrong!

Thanks

The expression on the right

IanH — Wed, 14 Jan 2015 20:55:54 GMT

The expression on the right hand side of the assignment to i is not using 64 bit integers.

Consider the difference between the print statements in the following:

  USE, INTRINSIC :: ISO_FORTRAN_ENV, ONLY: INT64
  IMPLICIT NONE
  PRINT *, 2**32-2
  PRINT *, 2_INT64**32-2
END

Thanks Ian. the problem is

Alireza_Forghani — Wed, 14 Jan 2015 21:20:58 GMT

Thanks Ian. the problem is fixed!