MRI: Acquisition of a High-Performance GPU Cluster for Research and Education

Graphics Processing Unit(GPU)-acceleration) has become a useful, if not essential, tool for computational tasks in support of investigations across a wide range of disciplines, from data science to biological sciences to physical and social-science areas. Therefore, the addition of GPU-accelerated computing capacity comprising a cluster of hybrid GPU/CPU compute nodes to the HPC system, Hawk at Montclair State University provides critical support for research and education activities, across the Institution, which share the need for this capacity. The impact of each of these investigations is magnified by users’ access to the GPU-accelerated computing capacity provided by this System. Moreover, by serving as a focal point for high-performance, GPU-accelerated computing, the system stimulates and supports collaboration across disciplines by investigators as well as their external collaborators. In the setting of the university, a designated Hispanic-Serving Institution with a diverse student body, PIs, as well as other major users utilize this instrument in both undergraduate and graduate education activities that enhance the STEM education for these students, including a significant proportion who are first-generation students in higher education. The instrument serves to both attract the interest of --and provide training in leading-edge computing to-- a diverse group of students in order to inspire and prepare them to be part of the future STEM workforce by providing `hands on’ access to GPU-accelerated HPC and the computing tools and techniques enabled by this platform.The cluster of hybrid GPU/CPU (Computer Processing Units) compute nodes acquired under this award are integrated to the Hawk system. These nodes constitute a substantial and transformative expansion of computing capacity for research and education at Montclair State, which previously had no generally accessible GPU-accelerated HPC. The utility of the cluster of compute nodes is significantly enhanced by leveraging the resources of the Hawk cluster. The System is configured to be accessible to not only PIs, but also to users across the Institution, as well as their external collaborators and supports multiple ongoing GPU-accelerated research activities including investigations in (1) computing and data science, (2) biology and genomics, (3) geomorphology and land use, (4) computational mathematics, (5) applied mathematics, (6) business analytics, and (7) linguistics. These collaborations include the implementation of new techniques and approaches (e.g., machine learning and deep neural networks) to both emerging and long-standing problems in these disciplines.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.