University receives grant to construct high-performance computing cluster
This funding will facilitate the creation of a new computational cluster, “Accelerating Computational Research for Engineering and Science” (ACRES), at Clarkson. It will provide computational resources to support current research and provide a structure for future growth.
A computing cluster is a group of independent computers most commonly linked together through a high-performance network.
ACRES will include 2440 Intel Xeon Cascade Lake cores, 8TB of memory, four NVIDIA Tesla V100 GPUs, 40Gbps Infinniband interconnect, and 40TB of SSD scratch storage.
Clarkson Chief Information Officer Joshua Fiske, principal investigator of the NSF award, said, “Access to high-performance computing resources is critical to expanding these research activities.”
As proposed to the NSF, the project will support ten active research projects in Clarkson’s four research focus areas: data and complex system analytics, healthy world solutions, advanced materials development, and next-generation healthcare technologies.
“ACRES will facilitate research currently not practical or feasible,” Fiske said. “It also will support new student learning opportunities through courses, undergraduate research, and an existing NSF research experience for undergraduates that focuses on high-performance computing.”
“ACRES will be available for use by all Clarkson faculty and their research teams to support sponsored research,” said Brian Helenbrook, professor and chair of mechanical & aeronautical engineering and the award’s co-principal investigator. “It will support the University’s increased focus on computational research as well as the hiring of additional computationally active faculty.”
For users who need more than casual access to a shared computing environment, Clarkson will also offer faculty members the option of purchasing additional dedicated resources to augment ACRES by becoming ACRES owners.
“We anticipate that when it’s fully operational the vast majority of ACRES’s compute nodes will be owners’ nodes, and researchers’ purchases will be the main driver behind the expansion of the cluster,” Fiske said. “This is often called ‘the condo model’ and allows ACRES owners to benefit from the scale of the cluster, giving them access to more compute nodes than their individual purchase. This provides owners with much greater flexibility than owning a standalone workstation.”
– Edited by Chris Vavra, associate editor, Control Engineering, CFE Media and Technology, firstname.lastname@example.org.