KNOXVILLE – Top scientists are lined up for a chance to get their hands on the Kraken supercomputer, but a group of undergraduate students from the University of Tennessee, Knoxville, had a unique opportunity to put the computer to use.
The computer, funded by a $65 million grant to UT Knoxville from the National Science Foundation (NSF), is the fastest academic supercomputer in the world, and the opportunity to use the machine is rare enough for elite researchers, much less for undergraduates.
More than 30 students in the Data Structures and Algorithm Analysis class in UT Knoxville’s department of electrical engineering and computer science (EECS) were given special user accounts on the computer, a Cray XT5, that allowed them to write and test different programs on the computer. One of the students, Chris Richardson, said the experience was a look at the forefront of science.
“The thing that opened my eyes the most was the way that this is going to be the wave of the future,” he explained. “There is always going to be a desire for faster and more efficient computing, and this will be it.”
The class, led by David Banks, an associate professor of electrical engineering and computer science with a joint appointment in the UT-Oak Ridge National Laboratory Joint Institute for Computational Science, had the opportunity to see the differences between programming on their own computers and on Kraken, which is roughly 100,000 times more powerful than the average home computer.
“JICS is a unique organization,” said Banks. “It is the only place in the world where faculty at a flagship university can partner with the staff at a national laboratory in order to teach undergraduates from different departments how to write code and run it on a world-class supercomputer.”
Before working on Kraken, the students were trained in how to make the computer’s 60,000 individual processors talk to one another effectively.
Staff scientists and faculty from the UT-ORNL Joint Institute for Computational Sciences taught the students the basics in their on-campus programming labs to ensure they were ready to use their allotted time on Kraken effectively. That preparation and training stands to serve the students well as they continue in their careers.
“Multi-processor programming definitely sets you apart from the competition when applying for any computer-related job,” said sophomore Michael Jugan, a computer engineering major from Knoxville. “Making a program run efficiently on multiple processors requires a better understanding of the computer’s hardware than regular coding does.”
Computational science is emerging as a vital element in how scientists conduct their work, using computers’ analytical power to address questions in ways that were not previously possible
The students used Kraken remotely, just like other scientists and researchers from around the U.S. who use the computer as part of the NSF TeraGrid network. John Cobb of ORNL sponsored the initial TeraGrid account for this project.
At the end of the semester, members of the class were given the opportunity to see the machine up close and meet with staff from the National Institute for Computational Science, the UT-led consortium that manages the computer.
“It is important for the nation’s continued leadership in computing that we expose as many students as possible, as early as possible, as often as possible, to programming on these high-end platforms,” said NICS Director Phil Andrews.
“When I was standing and looking at Kraken, I could not help but wonder how much data was passing through the machine at that moment,” said David Prenshaw, a student in the class.
Kevin Tomsovic, professor and head of EECS, said that the experience serves as a model way to inspire tomorrow’s computational leaders.
“Two great challenges in computer science education today are to excite a new generation about the diverse range of problems we study and to attract top students to our field,” he said. “Incorporating supercomputing early in the curriculum helps accomplish both.”
C O N T A C T :
Jay Mayfield, (865-974-9409, email@example.com)