New class will have students building, running their own supercomputer
November 27, 2012
When his computer architecture class gets to solving bottlenecks various applications tend to hit, Ethan Madden figures he will be in familiar territory.
After all, the junior in computer science from Newburgh spent most of the semester figuring out how to efficiently run scientific applications in such fields as environmental modeling and quantum mechanics. Madden was part of a team of Purdue undergraduates that just finished competing in the student supercomputing competition at SC12, the world’s largest supercomputing conference.
“I got a lot of real world experience that others in my classes won't have, which will help me with my schoolwork, as well as look much better on my resume,” Madden says.
Now, Purdue is offering a new scientific computing class that gives other students the opportunity to, like the student supercomputing team, build and operate their own mini supercomputer.
Besides earning three credits in the class, EAS 391, which is open to any undergraduate, students might end up on the team for the SC13 supercomputing conference in Denver or on teams for the International Supercomputing Conference in Germany.
Harrell says each student in the class will assemble their own supercomputer, install an operating system and other operational software, such as a job scheduler and message passing interface, and then work with the Weather Research and Forecasting Model (WRF for short), a scientific application used by both weather forecasters and researchers. The students will use WRF with real weather-related data for central Indiana and cull the data for information, for instance a rainfall projection for a particular day.
“It’s an introductory, broad course,” Harrell says. “You’re going to get a taste of everything.”
Plans are for Harrell and Howard to teach another section of the class in the fall, in which students would concentrate on the hardware and the battery of scientific applications used in the student cluster competitions.
For the spring class, ITaP is funding LittleFe clusters, a small — it can even be checked as airline luggage — inexpensive, yet fully functional cluster supercomputing system specifically designed for use in scientific computing education.
Harrell and Howard say that while teaching students scientific computing obviously contributes to Purdue’s educational mission, it also contributes to the University’s research. ITaP hires student workers to help maintain the community cluster supercomputers at Purdue. The ITaP-operated array includes some of the most powerful research supercomputers in the world, which are used by hundreds of Purdue faculty, staff and graduate students.