MRI: Acquisition of Hybrid CPU/GPU Nodes for the Interdisciplinary UMBC High Performance Computing Facility

点击翻译按钮获取中文摘要

Proposal #: 12-28778PI(s): Matthias K Gobbert; Erill, Ivan; Olano, Thomas M; Sparling, Lynn; Thorpe, IanInstitution: University of Maryland Baltimore CountyTitle: MRI/Acq.: Hybrid CPU/GPU Nodes for the Interdisciplinary UMBC High Performance Computing Facility Project Proposed:This project, acquiring an instrument based on advanced GPU/CPU computers with 360 TB high-performance storage and high-speed communication links, aims to enhance the current UMBC High Performance Computing Facility. The instrument enables research in emerging data-intensive high-performance computing topics. This multi-disciplinary and collaborative project involves faculty and students from biology, chemistry, computer science and engineering, electrical engineering, environmental systems, mathematics, physics, and statistics. The following current, as well as new research projects, benefit immediately from the availability of the enhanced high-performance computing instrument:- Parallel Algorithms for Numerical Simulations of Partial Differential Equations; - Application of GPU Processing to Atmospheric Chemistry and Dynamics Modeling;- Atmospheric Remote Sensing; etc.Broader Impacts:Research and educational initiatives should be highly impacted by the research enabled through the instrument. The research team, including female faculty members, has interactions with the Women in Science and Engineering (WISE) organization and the Center for Women in Technology (CWIT). Undergraduate students will be involved in the UMBC REU site on high performance computing. Furthermore, the instrumentation should contribute to attract students from under-represented groups into sciences. Presented are plans and initiatives to do so.