M.J. Oudshoorn, M.M. Fuad, and D. Deb (USA)
Automated distribution, Distributed Shared Memory, GRID computing, Agents, Middleware.
There is great potential if one could effectively harness the computation power of the plethora of low priced powerful yet under-utilized workstations connected by high-speed networks. In recent years, rapid improvement of advanced hardware and progress in networking technology is also making PC clusters a competitive platform for parallel computing. Traditionally, cluster programming has been performed with message passing libraries. As the performance, especially the latency, of the communication network improves, the expectation to support a shared memory programming model on clusters is reviving. In this paper, we put forward a system having both a distributed shared memory cluster and heterogeneous computers as system nodes which can support automatic distribution, migration, communication and coordination of multithreaded applications on a heterogeneous platform. This will be a creative application-driven experimental infrastructure that will explore challenging problems in distribution management and control, in middleware integration and deployment, in software architecture and software engineering.
Important Links:
Go Back
