MRI: Acquisition of 100TF Graphics Processor Laboratory for Multiscale/Multiphysics Modeling

Johns Hopkins University, Baltimore MD

Investigators

Abstract

Complex systems that involve a wide range of length and time scales and different physical processes are at the forefront of science and engineering. New computational hardware and algorithms are needed to address these challenging problems. The goal of this award is to acquire a compute cluster built with graphics processor units that have the potential to dramatically accelerate calculations. The cluster will allow NSF funded research teams to develop new multiscale/multiphysics algorithms and implement them for graphics processors. Applications to solids will include modeling and optimization of the mechanical properties of materials with heterogeneous structures, connecting atomistic dynamics to the macroscopic response of disordered metals and polymers, and fracture under conditions from slow laboratory loads to planetary impact. The facility will also enable studies of fluids that address temporal and spatial complexity in turbulent flows, water waves and air-sea interactions, and multiphase flow of fluids and particles in porous media. Every member of society is affected by decisions or predictions based on computational simulations of critical processes, such as infrastructural integrity, energy production, and environmental processes. This facility will provide a platform for the development and application of new modeling paradigms and techniques for such complex problems. The facility will serve as the primary "laboratory" for the NSF-Integrative Graduate Education and Research Traineeship program on "Modeling Complex Systems - the Scientific Basis of Coupling Multi-Physics Models at Different Scales." It will play an essential role in the training of the students and postdoctoral scholars associated with this program, producing a cadre of computational scientists who can lead the application of new methods for modeling complex systems.