A team of undergraduate students from the University of Alberta in Edmonton has won a major prize at the annual Supercomputing 2007 conference (SC07), held in Reno, Nev.
SC07 is the premier international conference on high-performance computing, networking, storage and analysis.
The University of Alberta team - the only Canadian team entered in the competition - beat four American universities and a university from?Taiwan district?in the first-ever "Cluster Challenge" held at the supercomputing conference. The five other teams were from the universities of Colorado and Indiana, Purdue University, Stony Brook University and the National Tsing Hua University in Taiwan.
The team from the University of Alberta consists of Antoine Filion, Paul Greidanus, Gordon Klok, Chris Kuethe, Andrew Nisbet and Stephen Portillo. All are undergraduate students at the university, except for Stephen Portillo, a 16-year-old student from Old Scona Academic High School in Edmonton. This talented high-school student, who took part in the U of A Department of Computing Science internship program, so impressed the department during his internship that he was asked to join the cluster team.
"We had a deep pool of student talent to draw from at the University of Alberta," said Paul Lu, professor in the Department of Computing Science. "Stephen was the top high school student on the first Iverson Computing Science Exam in June 2007. The students and their coaches worked very hard to prepare for the Cluster Challenge."
Along with Lu and co-coaches Bob Beck and Cam Macdonell (both postgraduate students in Computing Science), the team trained for weeks to assemble, configure and operate a cluster of powerful servers for scientific computation.
"Their hard work and preparation paid off when the entire convention centre lost power briefly on Tuesday, which required a complete restart of the cluster and the computations. I am proud of them all," Lu said.
The powerful cluster, provided by vendor partner SGI, has 64 processor cores and 144 GB of main memory. Most impressively, the entire cluster requires a relatively low 26 amps of current to operate - the low ampage was one of the rules of the Cluster Challenge. In fact, the Alberta team finished all of the computations for all of the applications in about 38 hours, some six hours before the official completion of the competition.
It is becoming common in science, manufacturing, film making, the financial industry and many other areas of the economy to use clusters. Clusters are powerful enough to tackle problems in seismic data processing for the oil and gas industry, computing financial risk, generating animated films and special effects as well as simulating the behaviour of chemical molecules and ocean currents.
This is the second time this month University of Alberta students have won in their category at an international competition. A multi-disciplinary team of students won first place in the energy category for their work on creating an environmentally friendly biofuel called butanol.
