September 16, 2005

Getting an HPC Education

By Peter Meade


Since its origin in February 1993, HPCwire has served as more than the preeminent location for news on the latest events in high performance computing. The publication has also assisted the HPC community as a meeting place for the exchange of concepts and ideas.

Frankly, some of HPCwire's most interesting original editorial contributions have been in response to features in previous issues. The latest example emerges this week as a trio of scientists from Purdue University's Rosen Center for Advanced Computing have written an insightful piece in response to an article from our Aug. 26 issue.

In the initial piece, “New Directions for Computational Science Education” (, authors Thomas Murphy, Paul Gray, Charlie Peck and David Joiner went into great detail on why they believe HPC education at the undergraduate level can be summed up in one word: broken. While I certainly do not disagree with their assessment, this week's contribution from Krishna Madhavan, Sebastien Goasguen and Gary Bertoline definitely propels the argument deeper and wider.

I only need to glance at my almost-five-year-old son to know today's educational system has moved forward at the speed of Blue Gene/L since his old man walked the halls of Glenwood Landing Elementary in the late 1960s. Yet Murphy & Co. and the Boilermaker trio both agree that much more needs to be done.

Arguably, the key point is that while undergrad HPC activities may be broken, the bigger picture is, sadly, so is the entire educational system. So much hope rests on the construction of a cyberinfrastructure that can enable educational science. As the “Let's Talk Xanga” feature in this issue states, education must accelerate to “where information technology provides the necessary foundation to build pedagogy.”

To create new, different ways of educating we need to start thinking and teaching in different ways. To think textbooks remain the main source of educational coverage is frightening. As Madhavan and his Purdue colleagues write, we “need to revise existing teaching methods to increase the focus on students.” For that rally cry, they deserve to move to the front of the class.

Let me take this opportunity to say HPCwire welcomes hearing from all its readers, the key movers and shakers in the high performance computing ecosystem. We want this publication to be a public form for airing and solving issues that are of major importance to your endeavors in HPC.

Please get in touch as I welcome your comments and contributions on what can be done to improve the educational experience for the current and next generations (as well as on other subjects). While I still have at least a dozen school years before my son heads off to Stanford or Berkeley, I have high hopes he won't be an English major like his parents (oh, please!) and that new wonders from all vistas of the educational experience will be awaiting him.

Peter Meade

Editor, HPCwire