MRI Consortium: Acquisition of a Heterogeneous, Shared, Computing Instrument to Enable Science and Computing Research by the Mass. Green High Performance Computing Consortium

The Massachusetts Green High Performance Computing Center, Inc., Holyoke MA

Investigators

Abstract

This work includes a consortium acquisition of a shared high-performance computing (HPC) instrument emphasizing HPC-based science and engineering with the benefit of "on-demand" capabilities. The instrument is a hybrid platform, integrating a state-of-the-art cluster computer with forward-looking, potentially exascale-oriented, GPU accelerator hardware that will enable progress in key STEM applications across a consortium of higher-educational partners with a research agenda that spans science disciplines. The system?s high levels of responsiveness will be applied to systems modeling and identification for immunology, interactive drug pathway analysis, bio-molecular electrostatics, virtual earth modeling and data analysis, multi-scale environment modeling, adaptive real-time model data synthesis, interactive virtualized materials design, and studying fundamental physics of matter. In computer science, researchers and industry collaborators will use the instrument as a controlled, configurable facility with an advanced user base and application workload, driving forward research into abstractions for next-generation HPC environments, enhanced virtualization methodologies for large-scale HPC, and quantitatively experimenting in cloud/service paradigms at scale for HPC. The acquisition will catalyze collaboration among scientists, engineers, computational science practitioners and computer scientists with research interests in - broadening the use of HPC for critical science and engineering problems - driving knowledge in earth science, life science, material science and basic physics - developing new, more engaging approaches to HPC - paving the way for a post-petascale HPC ecosystem spanning software to workforce skills - ere are extensive education and outreach activities planned to grow participation across institutional research communities as well as among local collaborators at community colleges, in K-12 education, and in community organizations. The instrument will contribute to cyberinfrastructure in multiple ways and will be a valuable part of a broader public-private partnership enhancing computational science infrastructure and catalyzing high-tech innovation.