MRI: Acquisition of a Graphics Processor Unit-Accelerated High Performance Computer for Astrophysics, Computer Science, and Broad Numerical Research at the University of Arizona

University Of Arizona, Tucson AZ

Investigators

Abstract

Due to the ongoing rapid increase in computing power, computer simulations and models of large numbers of particles (from atoms to stars) has become a powerful means of verifying theories of how large aggregates of matter form and evolve. It is now possible for literally trillions of objects to be followed in a computer as they are influenced by varying environmental conditions (gravity, pressure, density, temperature as well as the presence of dust, gas, strong magnetic fields, and even dark matter). These simulations show us how stars, planets, and galaxies may have formed, evolved, and died either quietly or in a violent explosion such as a Super Nova. These computer simulations are valuable tools for scientists and they also enable the computer programmers to produce "visualizations" (essentially movies, in 2- or 3-D) of these same phenomenon that are now widely used to show complex interactions on a computer monitor or projector screen to be enjoyed by all. Highly popular computer games and science fiction movies have driven the digital electronics industry to create "graphics processing units" that can very rapidly and efficiently manipulate digital pictures. Such GPUs are now common in smart phones, game consoles, and personal computers. On a much larger scale, scientists are now using GPUs in addition to the more traditional Central Processing Units (CPUs) of early computer days, to make large parallel processing machines (with hundreds of GPUs and CPUs) of very high power and speed to perform these complex calculations and manipulations. Dr. Brant Robertson of the University of Arizona is leading a team who will put together such a computer cluster to answer questions in theoretical astrophysics. The new high performance computer will also be available for general use by faculty and students at the University. Funding for this computer is provided by NSF's Division of Astronomical Sciences through the Major Research Instrumentation program.