ITaP activates new research supercomputing cluster, space still available
September 30, 2010
That rocket engines and pipe organs have something in common may not seem intuitive, but both blow acoustic waves through a tube. This can result in beautiful music or, in the case of an engine, operating inefficiencies, even catastrophe.
Charles Merkle, an aeronautics and astronautics and mechanical engineering professor, is interested in fluid flow and combustion aspects of rockets and gas turbines. He and his students examine such processes minutely by breaking them into myriad tiny steps played out over time in computer models.
ITaP’s new Rossmann supercomputing cluster is helping Merkle’s lab meet the considerable computational demands of the research. He’s gone from being able to enlist 50 processors a few years ago to 500 processing cores in Rossmann, the third high-performance research computing system ITaP has built in as many years.
“The three machines together have really changed what we do,” Merkle says.
Rossmann, part of ITaP’s award-winning Community Cluster Program, was assembled by ITaP staff and student employees in August and became operational in September.
Room for expansion remains in the new cluster, named for Michael Rossmann, Purdue’s Hanley Distinguished Professor of Biological Sciences and a pioneer in using high-performance computing to deduce the structure of viruses and their component protein molecules.
The Rossmann cluster features HP compute nodes with dual 12-core AMD “Magny-Cours” processors and 48, 96, or 192 gigabytes of memory. The new cluster has three times more processing cores and memory per node than in Steele and Coates — the clusters installed in 2008 and 2009. The new system also includes a high-performance Lustre parallel file system and 10 Gigabit Ethernet interconnects.
“In many ways what you get for your money is even more impressive than in the previous community clusters,” says Bill Whitson, director of research support at ITaP’s Rosen Center for Advanced Computing.
Physics Professor Norbert Neumeister says Rossmann’s price per processing core was particularly attractive to the Compact Muon Solenoid (CMS) data analysis center on campus. The center already has 400 terabytes of data to examine from the Large Hadron Collider, the giant international particle accelerator project probing the boundaries of our understanding of physics and the universe. A lot more data is coming.
The CMS project is running 1,000 jobs at a time on Rossmann now and should reach 3,000. “The more jobs you can submit, the faster you can analyze this data,” Neumeister says.
In addition to physics, aeronautics and astronautics and mechanical engineering, Rossmann has attracted faculty members from civil engineering, electrical and computer engineering, biology, chemistry and statistics, among other fields. Besides the CMS project, the Purdue-based Network for Computational Nanotechnology and nanoHUB.org, directed by electrical and computer engineering Professor Gerhard Klimeck, are major users.
Campus Technology magazine highlighted the Community Cluster Program in its August edition as one of the magazine’s 2010 international Campus Technology Innovators award winners. The magazine called Purdue’s program a model for providing university research computing services.
Through community clustering, ITaP pools internal and external funds to make more computing power available for Purdue research projects than faculty and campus units could afford individually.
ITaP installs, administers and maintains the community cluster systems, including security and user support, so researchers can concentrate on doing research rather than on running a high-performance computing system.
Community clustering also maximizes the use of resources by sharing computing power when it is idle. Researchers always have access to the capacity they purchase, and potentially more if they need it.
In addition, the program provides campus units, or individual researchers purchasing computing equipment for use outside Rossmann, an opportunity to leverage price breaks available in a group purchase.
Writer: Greg Kline, science and technology writer, Information Technology at Purdue (ITaP), 765-494-8167, firstname.lastname@example.org