This award provides support for a research infrastructure based on a cluster of high-performance, multiprocessor workstations connected via high-speed networks. It will enable new research activities in parallel and distributed computing systems and applications, more realistic and robust experimental studies, and generate synergy among the systems research and a number of different, but related, application research projects. Key systems research areas which the infrastructure will enable are cooperative optimization strategies, interprocedural optimization for distributed address space architectures, architectural support for multithreading on a cluster, load balancing and task scheduling, dynamic techniques for providing quality of service guarantees, and network management. Application areas include gesture recognition, multimodal applications, real-time multi-agent systems, video/image processing, and parallel symbolic analysis.
The current infrastructure consists of 20 workstations, connected by both 100Mbps switched ethernet and Myrinet.
The infrastructure provides both a dedicated system for applications researchers and a crashable resource for experimental parallel and distributed systems research.
The second year progress report on the research is also available in both PostScript and PDF.
The third year progress report on the research is also available in PDF.