compiler options

[abaraldi]

Hi,
I am compiling a large parallel code with intell 11.1 fortran compiler on a Linux cluster.
I am trying to port my code there from a version running on a CRAYXT5 and a Sun constellation cluster.
My problems are some options I cannot fnd anymore: basically what I need is
#standard double precision
#initialization to zero of variables and arrays
# -save option

about double precision, I fopund the option -dbl that works. I don't know why -r8 does not work.
for array initialization, the manual says -zero, but this is not regognized by the compiler, same for -save:

f95: unrecognized option '-zero'
f95: unrecognized option '-save'

Any hints anyone?
Thanks a lot
Alberto

[Steven_L_Intel1 · Accepted solution]

You're not using Intel Fortran for those compiles. If you were, the errors would not begin with "f95:".

[abaraldi]

Quoting - Steve Lionel (Intel)

You're not using Intel Fortran for those compiles. If you were, the errors would not begin with "f95:".

You are right. Thanks a lot. I am apparently using GNU fortran. The problem is that I want to use Intel, which is already installed, but I have to find a way to link it to the MPICH library.
I found a script under /sw/intel/Compiler/11.1/056/bin/intel64: ifortvars_intel64.sh
But this is just setting up the fortran environment. Any hints on how to link the fortran compiler to MPICH n a standard way?
Thanks a lot

[TimP]

The standard way to use ifort with an open source MPI is to build a new copy of that MPI in accordance with its instructions. The MPI fortran library has to be built with the Fortran you intend to use, and installed across your cluster, assuming you use shared library. openmpi is probably the most often used such MPI, although there must be at least half a dozen common variants of MPICH.
Another choice is Intel MPI, which comes pre-configured for both gfortran and ifort, as well as corresponding C and C++, and is available in evaluation and beta test versions. Intel MPI is derived from MPICH2.