LAM: LAM/MPI and Lahey Fortran - problem with mpirun

Hello,

after a few days of struggling and looking for the answer
I decided to ask here.

I installed LAM/MPI 7.1.3 on my Kubuntu 7.05.1 (Feisty
Fawn):

./configure --prefix=/opt/mpi --with-fc=lf95
make
make install

(Lahey Fortran 6.2-1 was already installed and working fine
and enviroment variables PATH, LD_LIBRARY_PATH set
properly).

Installation completeted successfully. After  compilation
of the code tmpi.f90:

!------------
program testmpi
implicit none
include 'mpif.h'

integer ierr
call MPI_INIT(ierr)
print*,"hello..."
call MPI_FINALIZE(ierr)

end program testmpi
!------------

with

mpif77 tmpi.f90 -o myprog
(no errors, no warnings, executable created)

I tried:

mpirun -np 2 ./myprog

and I got:

------------------------------------------------------------
-----------------
 It seems that [at least] one of the processes that was
started with
 mpirun did not invoke MPI_INIT before quitting (it is
possible that
 more than one process did not invoke MPI_INIT -- mpirun was
only
 notified of the first one, which was on node n0).

 mpirun can *only* be used with MPI programs (i.e., programs
that
 invoke MPI_INIT and MPI_FINALIZE). You can use the
"lamexec" program
 to run non-MPI programs over the lambooted nodes.

------------------------------------------------------------
-----------------

It seems that something really wrong happened, and the
executable
didn't even reach MPI_INIT. What could be the problem here?

What's more interesting, everything works fine if I use
gfortran instead of Lahey. And example programs in C cause
no problems in both of the cases (compiled with mpicc).

I also tried LAM/MPI test suite. All tests with C files
passed and failed on *.f files.

Any hints?


Best regards,
Maciej Marek 
_______________________________________________
This list is archived at http://www.l
am-mpi.org/MailArchives/lam/

On Sep 2, 2007, at 5:04 AM, Maciej Marek wrote:

> mpif77 tmpi.f90 -o myprog
> (no errors, no warnings, executable created)
>
> I tried:
>
> mpirun -np 2 ./myprog
>
> and I got:
>
>
------------------------------------------------------------
---------- 
> -------
>  It seems that [at least] one of the processes that was
started with
>  mpirun did not invoke MPI_INIT before quitting (it is
possible that
>  more than one process did not invoke MPI_INIT --
mpirun was only
>  notified of the first one, which was on node n0).
>
>  mpirun can *only* be used with MPI programs (i.e.,
programs that
>  invoke MPI_INIT and MPI_FINALIZE). You can use the
"lamexec" program
>  to run non-MPI programs over the lambooted nodes.
>
>
------------------------------------------------------------
---------- 
> -------
>
> It seems that something really wrong happened, and the
executable
> didn't even reach MPI_INIT. What could be the problem
here?

Can you run with just one process?  If not (and I'm guessing
not),  
can you try running under a debugger and seeing where the
failure is  
occurring?  As long as you've already run lamboot, you
should be able  
to do something like:

   gdb ./myprog

And run under a debugger to see where the failure occurs.


Brian

-- 
   Brian Barrett
   LAM/MPI Developer
   Make today a LAM/MPI day!


_______________________________________________
This list is archived at http://www.l
am-mpi.org/MailArchives/lam/