Computer Science Seminar Series
Parallel Programming with MPI
Friday, April 29th, 3:00pm
Weir Hall, Room 235
Presenter: Tyler Simon
Supercomputer Consultant, Mississippi Center for
Supercomputing Research
In the scientific community many sequential programs require hundreds or even thousands of hours to complete. Thus by dividing the computational bulk of a program over multiple processors (as in a Cluster) we can often see improved performance in regards to the number of total operations calculated per second. With this in mind we can use message passing routines within a distributed computing architecture to divide the sequential program into parts.
Most parallel programming continues to be done in either Fortran or C augmented with functions that perform message-passing between processes. The MPI (Message Passing Interface) standard is the most popular message passing specification supporting parallel programming and virtually every commercial parallel computer supports MPI.
This talk will present an introduction to MPI functions and demonstrate their use in sequential C programs. It will address some basics of parallel architectures, and provide higher level explanations of message passing over these. We will also learn some techniques for analyzing and developing parallel algorithms to be more performance oriented.
Last Updated: Wednesday, January 18, 2006, 11:59:06 CST