Scientific Computing - HPC Source November 2014

To help manage the hardware in the cluster, Mader leverages the capabilities of HP Integrated Lights-Out (iLO) software. “iLO integrates nicely with Bright Cluster Manager, which we are using to manage the cluster,” Mader says.

“Bright is fully aware of all the hardware and can manage it accordingly.” Mader considers the newly deployed cluster a first step on a path that will lead to a larger environment as demand for the center’s HPC resources grows among academic researchers, most of whom do work on government-funded contracts, and the automotive parts manufacturers who can’t afford the HPC resources and licenses on their own.

Click to read more

working collaboratively with NCMS to select the hardware configuration.

The cluster is built around an HP ProLiant DL380p head node and powered by 12 HP ProLiant SL230s Gen8 server trays with Intel® Xeon® processor E5-2670 in an HP ProLiant SL6500 scalable system. The cluster uses a TrueScale InfiniBand interconnect provided by Intel®. The HPC system’s primary application is LS-DYNA software from LSTC for crash simulations.

The center rolled out the 12-node system in 2013, after it moved to George Mason University from its longtime home at The George Washington University, says Jason Mader, the computer scientist who manages the IT environment for CCSA.

He notes that the center immediately got a 25% increase in per-core performance over its earlier system at The George Washington University, thanks to the use of next-generation Intel® Xeon® processor E5-2670 and improvements in code. The center’s previous three-year-old cluster was based on the Intel® Xeon® processor 5600 series.

This is a fitting role for the two centers, which have complementary goals. CCSA supports the U.S. Department of Transportation’s strategic goal to reduce fatalities and injuries on the nation’s roadways. NCMS, in turn, is the driving force behind a national strategy to make HPC resources available to small and midsize manufacturers across the country.

The combination of the resources and expertise of the two centers opens up new opportunities for SMB manufacturers who make many of the parts that go into the larger products created by global original equipment manufacturers (OEMs). They can now access the HPC resources and leading modeling and simulation applications to make their vehicle components safer and more reliable.

“Among small and medium-size companies, even those with 2,000 employees, just a sliver of them—3% or 4%—have even basic simulation tools, beyond desktop,” says Jon Riley. “Our mission is to look for avenues where we can introduce the technology, document the value, help them explore the return on investment, and then give them an enabling mechanism to affordably access these technologies and explore their benefits in real-life applications.” The focus of the initiative is very much on the applied research that will lead to better products, as opposed to more theoretical explorations.

“We’re not looking at basic research here. Small and medium-size companies are thinking day to day, week to week, month to month, maybe year to year. They are not looking out 10, 12 years. So we’re trying to work in a very applied way to get solutions into their hands,” says Jon Riley, Vice President for Digital Manufacturing, National Center for Manufacturing Sciences.