MRI: Acquisition of a High-Performance Computing System for Scientific Research and Education at NDSU

点击翻译按钮获取中文摘要

This award to North Dakota State University (NDSU) funds the acquisition and commissioning of a high-performance computing (HPC) instrument that will significantly expand and update the resources at NDSU and in the state of North Dakota for scientific research and education. The HPC system will provide a state-wide regional resource giving faculty and students of the North Dakota University System (NDUS) appropriate infrastructure to engage efficiently in challenging research that requires parallel computing. It will facilitate hands-on training in HPC techniques for large-scale compute- and data-intensive analyses. The new computing facility is essential to the growing spectrum of research and training activities and will be used to foster collaborative relationships with research and education partners within North Dakota as well as nationally and internationally. At NDSU and beyond, the new HPC system will enable cutting-edge research in multiple areas, including fluid dynamics, biomedical engineering, physics, chemistry, materials science and engineering, precision agriculture, plant sciences and plant pathology, artificial intelligence, health care, and financial and business analytics. Beyond NDSU, the new instrument will provide HPC resources to researchers and students at the tribal colleges (TCs) in the state of North Dakota and the primarily undergraduate institutions (PUIs) and Master’s colleges/universities (MCUs) within NDUS. Continuing present outreach activities is planned to stimulate interest in science and engineering within the state.The project will provide a new instrument that consists of a fast, tiered storage subsystem with a parallel file system and a distributed memory hybrid HPC cluster (CPUs, GPUs, big-memory nodes, and high speed interconnect) specifically designed to efficiently process massive amounts of data as well as handle compute-intensive applications. The compute nodes will be capable of a combined theoretical peak performance of 98 TFLOPS whilst using 64-bit double precision GPUs will lead to a combined theoretical peak of 140 TFLOPS. This gives a total peak performance of 238 TFLOPS for the compute cluster. The research projects undertaken by the research groups will contribute to multiple fields, including those listed above. Consequently, more than 15 NDSU lead computational researchers and their groups (100+ scientists and students) have joined in this project to use an extensible HPC instrument architected for enhanced support for modeling, data collection, generation, analysis, storage, provenance, curation, and sharing. The PIs of the project will integrate students into their research projects and incorporate HPC into several graduate and undergraduate courses. Training workshop series and internship opportunities to train students and research staff in computational research using HPC will also be provided and supported by the commissioning and operations of this instrument. This will provide an excellent opportunity to train the next generation of HPC systems specialists. This award by the Office of Advanced Cyberinfrastructure (OAC) is jointly funded with the Division of Materials Research (DMR), part of the Mathematical and Physical Sciences Directorate, and the Established Program to Stimulate Competitive Research (EPSCoR).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.