MPI Configuration Guidelines

The Message Passing Interface (MPI) is commonly used to handle the communications between tasks in parallel applications. There are two versions of MPI, MPI-1 and MPI-2. Two implementations of MPI-1 (LAM and MPICH) and two implementations of MPI-2 (OpenMPI and MPICH2) are supported. Individual sites may choose to support only a subset of these implementations, or none at all.

In the past, running MPI applications on the EGEE infrastructure required significant hand-tuning for each site. This was needed to compensate for sites that did or did not have a shared file system, location of the default scratch space, etc. The current configuration allows jobs to be more portable and allows the user more flexibility.

The increased portability and flexibility is achieved by working around hard-coded constraints from the RB and by off-loading much of the initialisation work to the mpi-start scripts. The mpi-start scripts are developed by the int.eu.grid project.

Using the mpi-start system requires the user to define a wrapper script and a set of hooks. The mpi-start system then handles most of the low-level details of running the MPI job on a particular site.

Documentation

• MPI functional test specifications:
  • for lcg-CE
  • for CREAM
• EGI MPI User Guide
• The official MPI EGEE-III wiki for site managers
• Alfieri, R.; Covati, R.; Tagliavini, E.; Integrare il Calcolo locale MPI nella farm INFN Grid (pdf), CCR Technical Report INFN/CCR-08/04, Dic 2008
• Gridstat and its customization for MPI start
• How to run an MPI job in Pi2S2 (with JDL attribute for "MPI granularity") - Sep 2008

and other links to less up to date pages:

• The TCG Working Group on MPI (started in 2008)
• The WLCG wiki on MPI: MPI Working Group
• Old: web pages of the previous MPI Working Group