The Message Passing Interface is a standard for passing data and other messages between running processes which may or may not be on a single computer. It is commonly used on computer clusters as a means by which a set of related processes can work together in parallel on one or more tasks. Unlike the SMP / OpenMP approaches to parallelism, the parallel strands of execution in a MPI environment do not share any memory: these strands (processes) must therefore communicate data and other information by passing messages between each other.
MPI is used on systems ranging from a few interconnected Raspberry Pi’s through to the UK’s national supercomputer, Archer.
The Message Passing Interface (MPI) itself is just a specification for a message passing library.
There are multiple implementations of this specification, each produced by a different organisation, including OpenMPI and MVAPICH. This documentation includes information on the MPI implementations available on ShARC and Iceberg. These implementations have been compiled on our clusters in a way that allows them to make optimal use of each cluster’s high-speed network infrastructure (OmniPath on ShARC, Infiniband on Iceberg). If you are not sure which implementation to use then try the latest available version of OpenMPI.
To use MPI you need to:
mpi on ShARC, openmpi-ib on Iceberg);module load to activate a particular MPI implementation (or module load an application that itself loads an MPI implementation behind the scenes).Here is an example that requests 4 slots (CPU cores) with 8GB of RAM per slot then runs a program called executable in the current directory using the OpenMPI library (version 2.0.1, built using version 6.2 of the gcc compiler). It is assumed that executable was previously compiled using that exact same MPI library. The Parallel Environment is specified using -pe.
#!/bin/bash
# Request 4 MPI 'slots' (cores)
#$ -pe mpi 4
# Request 8GB of RAM per slot
#$ -l rmem=8G
# Load a MPI library
module load mpi/openmpi/1.10.4/gcc-6.2
# Run a program previously compiled using that specific MPI library
mpirun ./executable
Some more example MPI jobs are available in the HPC Examples repository of Sheffield’s Research Software Engineering group
Our general-access interactive queues currently don’t have any MPI-compatible parallel environments enabled. Thus, it is not possible to run MPI jobs interactively.