Instrumentation Grant for Research in Parallel and Distributed Computing

点击翻译按钮获取中文摘要

EIA-9986032David K. LowenthalUniversity of GeorgiaInstrumentation Grant for Research in Parallel and Distributed ComputingThe Parallel and Distributed Computing Laboratory at the University of Georgia, established in 1998 is dedicated to making parallel computing simple, portable, and efficient. The activities of the laboratory include designing and implementing efficient parallel algorithms, developing highly configurable environments for the monitoring, visualization, and interactive steering of distributed programs, and developing lower-level systems that transparently and dynamically determine how to distribute data across processors. Currently research in the laboratory is primarily done on networks of commodity uniprocessor workstations. The attractive price/performance ration of small, bus-based multiprocessors relative to uniprocessors makes it likely that two-and four-way multiprocessors will be commonplace within a few years. Still, to create large parallel machines, these multiprocessors will be networked together to create a hybrid (both shared-and distributed-memory) machine. This poses interesting problems, including (1) should the machine be treated as distributed or both shared and distributed. (2) How do we make full use of such machines and (3) how can a user monitor and control program on such machines? The Department of computer Science at the University of Georgia seeks funds to purchase eight Pentium-based multiprocessors with four processors each. A 100Mb/s Ethernet and an SGI fronted to serve as a console for visualization and interactive control. This equipment will be used by three projects.Parallel and Distributed algorithms for computational biologyMonitoring, visualization, and interactive steering, andAn integrated compiler/run-time system for global data distribution