Hello Jeff,
thank you for your feedback.
I have this problem when I compile c, c++ and fortran code with the mpi
bindings, i.e. mpicc, mpicxx and mpif90. If I compile directly codes with
gcc, g++ or gfortran I do not have this problem. This is why I think it's
something related to openmpi.
Thanks!
> It looks like your PATH is pointing to the Intel MPI mpirun,
> not to the Open MPI mpirun/mpiexec.
[Tom]
Just to expand a little on this, it looks like your path is pointing to the
Intel MPI run-time version (mpirt) that is included with the Intel Compiler and
it's PATH/LD_LIBRARY_PATH is set
Hi
It looks like your PATH is pointing to the Intel MPI mpirun,
not to the Open MPI mpirun/mpiexec.
echo $PATH
may give you a clue of what is going on.
Make sure you use the same MPI implementation
to compile (mpicc, mpif90, etc)
and to launch the program (mpirun/mpiexec).
Check the OpenMPI FA
Pradeep Jha ccs.engg.nagoya-u.ac.jp> writes:
>
> Hello,
> When I am trying to run a perfectly running parallel code on a new Linux
machine, using the following command:
>
> --
>
> mpirun -np 16 name_of_executable
>
>
I think you cited the correct prior thread on the OMPI users list, and I think
it correctly identifies the problem: the linker is finding the "wrong" library
first, determines that it is wrong, and skips it.
You probably should remove the "wrong" library from your LD_LIBRARY_PATH (e.g.,
if you'
No. Open MPI has its own shared memory transport.
On May 9, 2013, at 3:34 AM, Bob Ilgner wrote:
> Dear Community,
>
>
> When processing on a shared memory processor, does open-MPI use the same
> "nemesis" channel approach as used by mpich ?
>
> Regards, bob
>
>
> ___
Dear Community,
When processing on a shared memory processor, does open-MPI use the same
"nemesis" channel approach as used by mpich ?
Regards, bob
