Why a hybrid MPI/OpenMP programming strategy?

This year, Moore’s Law turned 50. The law, formulated by Gordon E. Moore, co-founder of Intel, suggests that “The number of transistors incorporated in a chip will approximately double every 24 months”. Until now, the law has always been validated. But, since few years, depsite the fact that the law is still valid, the processors wall-clock frequency seems to stagnate, and the law reaching a limit, mainly due to prohibitive electrical consumption.

Which MPI Implementation?

Which MPI implementation to choose? Which one is better? Which one is more suitable for my application? What is the difference between MPICH, Intel MPI and OpenMPI?

Who didn’t ask one of these questions, at least when he / she first started using or developing programs on HPC systems?

Install MPICH On Linux

In a previous post, we have detailed how to install OpenMPI on a Mac OS X system. At that time, it seemed alot easier to setup the OpenMPI implementation on a Mac system, rather that MPICH.

In this post, we will try to describe how to setup MPICH on a Linux system.

First, if not done yet, please consider installing the libc-dev package, which contains headers for developers from the Linux kernel: