CISE Research Instrumentation: A High-Performance Network for Distributed Computation and Visualization

9617541 Farrell，Paul A Ruttan，Arden G Ruttan州立大学Cise研究仪器：分布式计算和可视化的高性能网络仪器赠款使人们能够购买连接高性能的高带宽、低延迟光纤通道网络，用于以下研究项目：-液晶计算的路径跟踪、可视化和转向-分布式操作系统研究-大型可伸缩并行量子蒙特卡洛该设备将支持几个研究项目，涉及分布式大型科学计算、计算转向和可视化方面的工作，以及分布式操作系统方面的工作。除了研究项目本身，计划将光纤通道上的算法和方法与ATM网络上的算法和方法进行比较。关于液晶计算的路径跟踪、可视化和转向的项目包括开发高效的算法来解决分布式网络环境中的大规模液晶问题，以及可视化和引导结果计算的工具。大规模可伸缩并行量子蒙特卡罗方法的项目包括研究高度可伸缩的电子结构计算实现中的通信和负载平衡。这些算法和性能将在高带宽网络和并行机上进行比较。关于分布式操作系统的项目是为了能够有效地利用多个工作站。这涉及用于异构机网络的基于微内核的操作系统，以及用于广域网和局域网的异构型分布式调度系统。

9617541 Farrell, Paul A Ruttan, Arden G Kent State University CISE Research Instrumentation: A High Performance Network for Distributed Computation and Visualization The instrumentation grant enables the purchase of a high bandwidth, low latency Fibre Channel network connecting high performance to be used in the following research projects: - Path Following, Visualization and Steering of Liquid Crystal Calculations - Distributed Operating Systems Research - Large Scalable Parallel Quantum Monte Carlo The equipment will support several research projects, involving work in distributed large scale scientific computation, computational steering and visualization, together with work on distributed operating systems. In addition to the research projects themselves, it is planned to compare the effectiveness of the algorithms and methods on Fibre Channel to that on an ATM network. The project on path following, visualization and steering of liquid crystal calculations involves the development of efficient algorithms to solve large scale liquid crystal problems in a distributed network environment and the tools to visualize and steer the resulting calculations. The project on large scalable parallel quantum Monte Carlo methods involves investigation of communication and load balancing in highly scalable implementations of electronic structure calculations. The algorithms and performance will be compared on high-bandwidth networks and parallel machines. The project on distributed operating systems is designed to enable the effective use of ensembles of workstations. This involves microkernel-based operating systems for networks of heterogeneous machines and heterogeneous distributed scheduling systems for WANs and LANs.