MPI techniques
The Message Passing Interface (MPI) is a standard developed by the Message Passing Interface Forum (MPIF) for writing parallel programs. MPI programs are in particular well-suited for running on distributed memory machines, such as Linux clusters.
There are a number of MPI courses available on the Internet. However, most of these teach people to write hello-world style programs. For scientists the obvious question then becomes: how do I apply the knowledge from these hello-world programs to my research? This question can be difficult to answer, because there apparently is no connection between hello-world style of programs and advanced research projects. The missing link is "parallel algorithms". This section therefore focuses on: parallel algorithms implemented with MPI. Basic information, such as an explanation of the arguments used when sending or receiving data, will not be given here. The reader should use an Internet search engine, such as Google to look-up this information.
The information, figures and examples on these pages have with IBM's permission been copied from chapter 4 of IBM's Redbook: RS/6000 SP: Practical MPI Programming. However, several of the examples have been revised and updated. We thank IBM for allowing us to use their material in this way.
