A publication of the Office of Advanced Simulation & Computing, NA-114, NNSA Defense Programs
December 2007
NA-ASC-500-07—Issue
Return to this issue’s stories
SC07 Cluster Challenge Committee Proclaims: Perceived Entry Barrier to Supercomputing Has Dropped Significantly
At Supercomputing 2007 (SC07), six teams of undergraduates, supported by their chosen vendor partners, competed in a test of performance on the conference floor. Teams were given a few simple rules (26 amps and no one on the team can have a degree) and the summer to design the system that would perform best on a workload consisting of Linpack benchmarks (the industry standard High Performance Computing [HPC] Challenge) and three open source applications: GAMESS, a molecular chemistry application from AmesLab; POP, an ocean circulation model from Los Alamos; and POVRay, a popular ray tracing application.
Left: Cluster Challenge being introduced prior to event.
Cluster Challenge Committee Chair Brent Gorda, of Lawrence Livermore National Laboratory, observed, “The results showed that if you have a need for simulation computing, it’s reasonable to expect that you can use local college or university talent and commonly available software and applications to get started on that work.”
Six teams of undergraduates participated from the U.S., Canada, and Taiwan. The highest Linpack score of 420 gigaFLOPS was announced to cheers and applause at the Top500 Birds of a Feather (BOF). That performance would have made the list of the Top500 supercomputers just three years ago. For this to be done by a half-dozen undergraduates shows that there is no significant barrier for entry-level HPC. Conventional wisdom suggested that any system on the Top500 had to be backed by an institution, yet the challenge proved that a team of undergraduates could make the list.
The overall winning team was from the University of Alberta, in Edmonton, Canada, combined with SGI. The other competing teams were Stony Brook University and Dell, National Tsing Hua University (Taiwan) and ASUSTek, University of Colorado and Aspen Systems, Indiana University and Apple, and Purdue University and HP.
DOE Privacy Disclaimer | Sandia Privacy Disclaimer | SAND 2007-8167 W
Developed and maintained by Sandia National Laboratories for NA.114