ENGINEERING

The University of Texas at Austin team, mentored by staff of the Texas Advanced Computing Center (TACC), won the seventh annual Student Cluster Competition (SCC) this year at the Supercomputing '12 (SC12) conference in Salt Lake City, Utah.

SC12 has been the seminal international conference for the past 24 years.

The SCC competition challenges rival teams of university undergraduates in a 72-hour-battle to prove that they can design, build, optimize and run the fastest and most efficient cluster computing system. In this real-time, non-stop competition, teams of six students assemble their clusters on the exhibit floor and race to demonstrate the greatest sustained performance across a series of applications and scientific workloads.

The University of Texas team took first place in the overall standings, breaking Team Taiwan's (National Tsing Hua University) two-year winning streak.

"None of us expected to win, but we didn't expect to lose either…we were thrilled," said Craig Yeh, a third year Computer Science major at The University of Texas at Austin. "The win validated all of the work we did since May leading up to the competition. I highly recommend this experience to other students at The University of Texas."

The winning student team members are Andrew Wiley, Reid Douglas McKenzie, Michael Teng, Anant Rathi, Craig Yeh, and Julian Michael. Learn more about the team members: http://www.tacc.utexas.edu/sc12/student-cluster)

The competition draws teams from around the world, including the United States, Europe, Canada, China, Costa Rica, Germany, Russia and Taiwan. The teams work with company sponsors several months in advance to design and build a cutting-edge system from commercially available components.

This year, The University of Texas team and TACC partnered with Dell, Nvidia and Intel to design and build a hybrid, power-saving system that integrated graphic processing units (GPUs), Intel processors, and a new-generation Dell chassis. The company sponsors provide the hardware and travel funds, while the TACC mentors work side by side with the students to teach them the fundamentals of cluster construction, systems administration, and program optimization. Chevron and Mellanox also served as sponsors of this year's team.

One additional challenge of the competition is that the systems cannot exceed a 26 amp power limit, which is the electrical equivalent of three standard-size coffee-makers.

"It's a real-world situation," said John Lockman, the team's lead mentor and a member of TACC's High Performance Computing group. "For example, a data center in industry might need to expand, but can't due to financial or space constraints, so they have a limited amount of power and a scientific workload that they have to accomplish in a reasonable amount of time." John Cazes and Carlos Rosales-Fernandez, also from TACC, helped mentor the students.

The competition began early Monday morning, November 13, when the teams ran the HPCC benchmark, which tests multiple attributes that contribute substantially to the real-world performance of HPC systems. These results are factored into their overall score. They ran the Linpack benchmark independently to compete for the 'Highest Linpack' award (this famous benchmark measures the performance of the system solving a large number of linear equations). On Monday evening, all of the teams received the data sets for the four scientific applications that they ran over the next 48 hours.

HPCC benchmark scores, results from the scientific applications, and interviews with judges were weighted to determine each team's final score. Teams were also judged on the presentation of their system, their live visualizations, and how thoroughly they answered questions from conference participants.

TACC and its industry partners have sponsored student teams in the competition since 2010, when The University of Texas team received the highest Linpack score and was the first team in the competition's history to submit a Linpack score higher than a teraflop. In 2011, The University of Texas team built an energy-saving cluster that was submerged in mineral oil, which had the highest Linpack score among CPU-only teams.

"We had the opportunity to manage an entire project at a high level to get real science results. For me, the exposure to the scientific applications was the best part," Yeh said.