MRI/MII: Acquisition of Collaborative High Performance Computing and Visualization Cluster

stephen A. ZahorianR。该提案来自MII的一个机构，旨在建立一个用于高性能计算（HPC）和可视化的贝奥武夫集群，旨在实现、支持和加强建模和仿真方面的研究和教育。解决的主要应用程序使用各种数字密集型算法，这些算法需要多次“运行”基本计算算法才能查看全局。将从基础设施中受益的技术领域包括企业建模、仿真和可视化、PDE求解器和生物信息学等领域的高性能计算、控制系统建模、自动语音识别和物理电子模拟。ODU目前的具体研究活动包括：1。建模与仿真研究，2。国际空间站空间实验室窗口组件的JSC增强现实训练器，3。培训的虚拟环境，4。VMASC战斗实验室（战争游戏、人机界面、可视化和分布式仿真）计算科学中心，6；Terascale优化PDE仿真，7。生物信息学,8。语音通信研究，9。物理电子学模拟（生物电子学、纳米电子学、晶格玻尔兹曼模拟等）。生命关键控制系统研究。计算集群将在主校区有大约50个节点，在萨福克的建模和仿真中心也有一些类似的节点。通过千兆网络耦合并在两个子集群之间松散耦合，每个本地集群中的机器将使用公共领域（例如，基于Linux的）操作系统和调度，以便可以从不同位置启动计算密集型调查。该设备将支持教师和学生培训，提供先进的计算研究实践经验。

EIA-0216541Stephen A. ZahorianR. Bowen LoftinOld Dominion UniversityMRI/MII: Acquisition of Collaborative High Performance Computing and Visualization CenterThis proposal from an MII institution, establishing a Beowulf cluster for High Performance Computing (HPC) and visualization, aims at enabling, supporting, and enhancing research and education in modeling and simulation. The main applications addressed use a variety of numerically intensive algorithms, which require many "runs" of a base computational algorithm in order to view the big picture. Technical areas which will benefit from the infrastructure include enterprise modeling and simulation and visualization, high performance computing in areas such as PDE solvers and bioinformatics, control systems modeling, automatic speech recognition, and physical electronics simulations. Current specific research activities at ODU include:1. Modeling and Simulation Research,2. A JSC Augmented Reality Trainer for International Space Station Space Lab Window Assembly,3. Virtual Environments for Training,4. VMASC Battle Lab (war gaming, human computer interfaces, visualization and distributed simulation)5. Center for Computational Science,6. Terascale Optimal PDE Simulations,7. Bioinformatics,8. Speech Communications Research,9. Physical electronics simulations (Bio-electrics, Nanoelectronics, Lattice Boltzmann Simulations, and10. Systems Research on Life Critical Control Systems.The computational cluster will have approximately 50 nodes on the main campus and a number of similar notes at the modeling and simulation center in Suffolk. Coupled via gigabit network and loosely coupled between the two subclusters, the machines within each local cluster will use public domain (e.g., Linux based) operating systems and scheduling such that computationally intensive investigations can be initiated from various locations. The equipment will support faculty and student training providing hands-on experience in advanced computing research.