A specification for interprocess communication defined by the MPI-Forum.
====== MPI ======

There are no test or default implementations to guarantee the compatibility of different MPI-implementations.

The most common implementations are OpenMPI and MPICH, both incompatible to each other.

Open MPI 1.2 supports all of MPI-2.0

Open MPI 1.3 supports all of MPI-2.1

Boost.MPI supports all of MPI-1.1
Provided Versions on WW8-Clients
OpenMPI 1.6.4 (experimental)


  /usr/net/lib/openmpi-intel:

Intel compiler (sandy bridge optimized) with MKL

Usable after   

  source /usr/net/bin/module_intel

  /usr/net/lib/openmpi-gnu 
  
  
==== Supporting tools ====

[[tools:optimization | hwloc-ls]] Affinity Mask calculator

==== Fortran parameters ====
To get the paramters used py the MPI Fortran wrapers (mpif90)
<code bash>
mpif90 --showme
</code>
An example output of Fedora 23 with Openmpi (mpi/openmpi-x86_64) will result to:
<code bash>gfortran -I/usr/include/openmpi-x86_64 -pthread -I/usr/lib64/openmpi/lib -Wl,-rpath -Wl,/usr/lib64/openmpi/lib -Wl,--enable-new-dtags -L/usr/lib64/openmpi/lib -lmpi_usempif08 -lmpi_usempi_ignore_tkr -lmpi_mpifh -lmpi
</code>

==== Run job on multiple nodes ====
<code bash>
$(which mpirun) --hostfile [HOSTFILE] -np [NUMBER PROCESSORS] -x LD_LIBRARY_PATH -x PATH [EXECUTABLE]
</code>