On Jul 22, 2009, at 1:37 PM, Jeff Squyres (jsquyres) wrote:
> Yep, that works.
I should clarify -- that *probably* works.
The .mod file are essentially precompiled headers. Assuming that all
the data types and sizes are the same between gfortran and ifort, you
should be ok. Many of OMPI's F90 functions are implemented by
directly calling the back-end F77 functions, but some of them have
thin F90 wrappers before calling the back-end F77 functions.
If the calling conventions, parameter sizes, and constant values (see
that README that I cited earlier in this thread) are all the same,
then you should be ok using a single back-end libmpi_f77 and
libmpi_f90 with 2 different .mod files. But this is not something I
have tested extensively, so I can't give you a definite "this will
always work" ruling.
I *think* that there are compiler flags that you can use with ifort to
make it behave similarly to gfortran in terms of sizes and constant
values, etc. These may or may not be necessary...? You might want to
check into this, but I'd be interested to hear what you find.