f90 vs ifort

[crkkos]

I have in a code the lines

on real underflow ignore

on double precision underflow ignore

This worked with f90 before, but using ifort the code doesnt even compile saying

error #5082: Syntax error, found END-OF-STATEMENT when expecting one of: ( % : . = =>

on double precision underflow ignore

Is there a way to get this to work, or mimic this functionality with ifort?

[Ron_Green]

Quoting crkkos

I have in a code the lines

on real underflow ignore

on double precision underflow ignore

This worked with f90 before, but using ifort the code doesnt even compile saying

error #5082: Syntax error, found END-OF-STATEMENT when expecting one of: ( % : . = =>

on double precision underflow ignore

Is there a way to get this to work, or mimic this functionality with ifort?

Those statements are not legal Fortran, they are what are known as 'vendor extensions'. You'll need to comment out those lines. Then look in your documentation for the -fpe option to set various floating point exception handling modes. The default without -fpe is to disable all floating point execeptions.

But I think what you want is -no-ftz to allow denormals.

[crkkos]

i tried the -no-ftz and that in combination with fpe3. also tried various levels of optimization, but no luck =(

[TimP]

Ron didn't mean those options as a substitute for commenting out the non-standard syntax.

[crkkos]

i know, just saying sucks that this code used these old directives in the first place.