Utah State University’s Center for High Performance Computing has been selected as one of four finalists in the prestigious national Supercomputing Conference SC09 Storage Challenge, a competition showcasing real-world effective approaches using computer storage, which is often the main limitation of overall system or application performance.

HPC@USU will compete against teams from IBM research, San Diego Supercomputing Center, and a joint team from Microsoft Research, John Hopkins University, George Washington University, and visual computing technology leader NVIDIA. Finalists will present their projects November 14–20 at the SC09 Supercomputing Annual Conference in Portland, Ore. The winner will be announced November 20.

The HPC@USU team, lead by center director Thomas Hauser, includes mechanical engineering graduate student Kyle Horne and HPC@USU lead system administrator Nate Benson.

“The fact that Kyle, a student researcher, is included on the team speaks highly of his skills, and the talent of HPC@USU compared to other larger, better funded, institutions with deep industry support,” said Hauser.  “It really is a ‘David and Goliath’ scenario.”

HPC@USU has attended the international SC conference with student research exhibits since 2005, when HPC@USU was newly founded. Since SC06, HPC@USU has collaborated with the University of Utah Center for High Performance Computing to present a unified Utah HPC presence.

“I’ve always felt strongly about including student researchers in our activities,” said Hauser. “Not only have they presented research, but they have also been exposed to cutting-edge HPC technology, industry direction, and technical progress. This exposure helps inform their research at USU, as well as offer incredible networking opportunities for them on a national scale.”

One previous doctorate student of Hauser found his current job while at SC06, and all the student attendees have expanded their career opportunities (as well as the reputation of HPC@USU) through their participation, said Hauser.

HPC@USU’s SC09 project makes it easier to use parallel storage systems, which are designed to store and retrieve data faster than traditional systems. Previously, computational analyses were the most time-consuming parts of HPC research. Now, parallelism has reduced that processing time, so data storage and retrieval has become the bottleneck. Designing software to take advantage of faster storage systems will improve time-to-solution.

“The challenge is that parallel programming is hard,” said Hauser.  “HPC@USU’s software library simplifies it, making it more accessible to programmers.  That means that one doesn’t need to be a parallel programming or storage expert to take advantage of the capabilities of parallel storage systems.”

Competitors in the SC09 Storage Challenge will describe their implementations and present measurements of performance, scalability, and storage subsystem utilization. Judging will be based on these measurements, as well as innovation and effectiveness. The HPC@USU team is currently refining their implementation on a new HP cluster system connected to a Panasas parallel storage unit, and they are preparing their paper for the November conference.

“Our success shows that you can compete against the ‘big guys’ with a small team of talented and hard working members,” said Horne.

“HPC@USU is grateful for research funding from USDA, and the engineering support of Panasas,” said Hauser.

HPC@USU is a research support center at USU designed to enable research activities and education in all disciplines that depend on advanced computing technologies.  It provides large scale computing, visualization and data management resources; enables education of advanced cutting-edge simulation and cyber environments; and creates partnerships to develop and deploy new computing systems to advance education and knowledge. For more information, visit http://www.hpc.usu.edu.

The SC Conference series, held since 1988, is the largest international supercomputing conference and is sponsored by the Association for Computing Machinery and the Institute of Electrical and Electronic Engineers. SC occurs annually in a different U.S. city.  For more information, visit http://sc09.supercomputing.org/index.php.

Writer:             Barbara Sidwell, 435-797-1182  barbara.sidwell@usu.edu

Contact:          Thomas Hauser, 435-797-2834  thomas.hauser@usu.edu

Comments

Insert Comments

* Name:

* Email Address:

URL:

* Comments:

* Validation:

(Please enter the letters and numbers that appear below)