Researchers invited to participate in second high-performance computing cluster
1:46 p.m., April 28, 2014--Building on the success of the Mills High-Performance Computing (HPC) cluster, the University of Delaware is deploying a second community cluster to perform complex computational tasks for researchers in engineering; physical, natural, social, policy and decision sciences; and finance.
The new cluster will give more UD researchers access to HPC resources and will provide faculty with faster compute nodes and more storage than is available on Mills.
Snow monitoring network
The University uses community-cluster architecture for HPC systems. Periodically, IT staff work with University faculty to develop the specifications for an HPC cluster, with the purchase financed both by IT (cluster infrastructure, storage, and networking) and by individual faculty researchers (individual compute nodes).
The community cluster model is advantageous for researchers, allowing them priority access to high performance computing power while sparing them the ongoing financial liability of purchasing and running their own computing clusters.
"The Mills cluster has been one of the reasons why I accepted my position at UD," said Cristina Archer, associate professor in the College of Earth, Ocean, and Environment.
"Without it, I would not have been able to perform the computer-intensive simulations that I need for my research on turbulence generated by wind turbines. Being the biggest users of the cluster, my team and I have been interacting very closely with the IT team, whose support and help have been invaluable," Archer said.
Edward Ratledge, director of the Center for Applied Demography and Survey Research (CADSR) at UD, and his research group have also benefited from the availability of High Performance Computing power.
David Racca, policy scientist in CADSR and principal investigator for the project, said that the center has developed "a statewide travel speed survey to be used as transportation performance measures by transportation and land use planners and analysts."
"CADSR has collected a large amount information as provided by GPS readings from state vehicles over the last three years. On the fastest personal computer available, processing a month’s worth of this information can take over three weeks of continuous computer processing time, which proves to be a practically impossible task. The HPC cluster now processes a month’s worth of information in three days, making the development of this information possible and an ongoing statewide travel speed survey a reality," Racca said.
Compute nodes are still available for purchase in UD’s second HPC cluster. See IT’s information about the new cluster for details.
For any questions or to inquire about purchasing compute nodes, send email to email@example.com.