Teams prepare for SC13 Student Cluster Competition
11 November 2013Tweet
Eight teams from universities in the United States, Germany, China and Australia are preparing for the Student Cluster Competition taking place at SC13 on 17-22 November 2013, in Denver, Colorado, USA.
The participating teams are IVEC, a joint venture between CSIRO, Curtin University, Edith Cowan University, Murdoch University and the University of Western Australia (Australia); Massachusetts Green High Performance Computing Center (USA); National University of Defense Technology (China); The University of Colorado, Boulder (USA); The University of the Pacific (USA); The University of Tennessee, Knoxville (USA); The University of Texas, Austin (USA) and Friedrich-Alexander University of Erlangen-Nuremberg (Germany).
For the third year in a row, Silicon Mechanics will be sponsoring the Massachusetts Green High Performance Computing Center (MGHPCC). After identifying hardware partners and securing their commitment, Silicon Mechanics assembles and tests the cluster, and then ships it to the consortium for use in the competition. Valued at more than $118,000, this year’s cluster includes hardware provided by partners Mellanox, Supermicro, Intel, Kingston, Nvidia, and sTec. The HPC cluster contains eight compute nodes, four GPU nodes, and InfiniBand and Gigabit Ethernet networking.
The six-member MGHPCC team, which includes students from Boston University, Harvard University, MIT, University of Massachusetts, and Northeastern University, will compete in a 48-hour challenge to demonstrate the best sustained performance on four pre-determined applications. While they are given the application areas in advance to familiarise themselves, students will not know the specific tasks until the challenge begins.
This year’s MGHPCC team, comprised of computer engineering, electrical engineering, computer science, and physics majors, has already been busy practicing and optimising the contest applications. Competition application areas include WRF (Weather Research and Forecasting), a mesocale numerical weather prediction system; NEMO5, a multipurpose, multiscale, highly parallelised nanodevice simulation tool; GraphLab, a parallel computation abstraction tailored to machine learning; and a ‘mystery application’, which will be revealed at the competition.
The winning team will be determined based on a combined score for workload completed, benchmark performance, conference attendance, and interviews. Recognition will also be given for the highest Linpack benchmark. The awards for the Student Cluster Competition will be announced at the SC13 Awards Ceremony on Thursday, 21 November 2013.