A problem with double precision variables

[inkant]

Hi Ifort users,

I have recently com across this new (or perhaps old )strange error

forrtl: info (299): -1080016142 floating divide-by-zero traps. It does not take place with real varables.
forrtl: info (300): 1067264965 floating underflow traps

I could understand this in an exponentiation A**b but it is occuring with addition. Only in a particular code.
I am using Ifort 11.0.074 version.

I can assure you that there are no 0's involved in division directly. I am not sure how compiler would do a division.

any ideas?

[Steven_L_Intel1]

Do you have a STOP statement in the program? Fortran 2003 requires that when STOP is executed that the existence of unhandled floating point exceptions be displayed. These exceptions may have occurred in the math library. I do find the numbers shown rather suspicious, though. Is this a large program? Something you'd be willing to share with us?

[inkant]

suface_tension_new.f

Do you have a STOP statement in the program? Fortran 2003 requires that when STOP is executed that the existence of unhandled floating point exceptions be displayed. These exceptions may have occurred in the math library. I do find the numbers shown rather suspicious, though. Is this a large program? Something you'd be willing to share with us?

Hello Steve

There is no stop statement.

I added the file as an attachment.

The variables Coef_a, Coef_b, and Coef_c are now defined as DP vars. Just by adding "D0" at the end, makes them DP and should cause the problem for you also.

Best regards

[Steven_L_Intel1 · Accepted solution]

Try compiling with -CB

You declare dsig_DT and dsig_DYF as (11), but index them as 0 to 100 in the DO loop in the main program. This is bound (!) to cause problems.