April 30: WSU’s research computing to go condo

By Eric Sorensen, WSU science writer

PULLMAN, Wash. – With the acquisition this month of a $1.2 million high-performance computer, Washington State University’s research computing is going condo.

The new system, which will be available to WSU researchers across the state, is part of a growing trend in condominium cluster computing in which researchers own pieces of a larger computing system. They can use their share when they want, but when they are not using it, it is available to other university researchers.

“We are developing a community-based approach in which there’s continuous refreshing of the machine,” said Christian Mailhiot, a professor in the College of Arts and Sciences. “New users, instead of purchasing their own stand-alone system, are invited to contribute to a larger infrastructure.”

The computing system is part of a strategic plan for advanced research computing that Chris Keane, vice president for research, and Daryll DeWald, dean of the college, last year asked Mailhiot to lead. He will host a town hall discussion of the system at 4 p.m. Thursday, April 30, in CUE 203. A reception will be at 3:30 p.m.

At its peak performance, the new Dell system will run at approximately 20 teraflops, performing 20 trillion mathematical calculations per second. The condominium computer offers an economy of scale, said Mailhiot, while streamlining infrastructure and administrative needs. It will also be easy to expand.

Perhaps most important, it will have “surge” capabilities that will let researchers perform some tasks with speed and power well beyond what one can get in an isolated, stand-alone computer cluster. It was supported with a sizeable contribution from the College of Agricultural, Human and Natural Resource Sciences. In conjunction with a National Science Foundation-funded high-speed research network acquired last year, it will make possible large numerical simulations and data analysis used in fields like bioinformatics and genomics.

“It enables researchers to accelerate the analysis of large data sets,” said Mailhiot.

Read an earlier article about the computer system here.