High Performance Computing and Visualization SSS: Screen Saver Science

The Screen Saver Science (SSS) project aims to develop a computing resource composed of a heterogeneous set of PCs, scientific workstations, and other available computers, that can be easily used by scientists to execute large, highly distributed, compute-intensive applications. Each computer would participate only when it would otherwise be idle, that is, when its screen saver would be running. This work is based on Jini, a Java-based open software architecture intended for the development of robust network services. This project has several goals. First, we hope to utilize the idle processing power of the many PCs and workstations available at NIST to execute production scientific codes. The power of personal PCs and workstations continues to increase, and they are now becoming very capable of executing large-scale applications. Second, research on grid computing has been accelerating, and the SSS environment will provide a platform for local experimentation with highly parallel and distributed algorithms suitable for grid environments. Finally, the development of SSS applications will give us the opportunity to explore the use of Java on scientific applications, a topic of independent interest. Up until recently, a project of this scope would have required a large amount of software development just to become minimally functional, and hence was not practical. But, with the introduction of Jini, and its network service called Javaspaces, the most difficult parts of this project have now become trivial. Javaspaces is a portable, machine independent, shared memory system that expands upon the tuple-space concepts developed in the 1980’s by David Gelernter of Yale University. It provides the necessary infrastructure to support the development and use of remote services in applications. The SSS project began in the summer of 2002. Substantial progress has been made in designing and implementing the basic SSS infrastructure. A generic compute server has been implemented, suitable for embedding in a screen saver, and a new Jini service, a remote file server, has been developed to provide SSS applications with remote file I/O capabilities. With the basic SSS infrastructure now 90% complete, we have begun to develop the first large application for SSS, a quantum Monte Carlo application in ab-initio chemistry. This application is currently running on our Linux cluster but does not require the use of the cluster’s communications network since each task is completely independent. This makes it ideal for porting to the SSS environment. W. George presented the preliminary design of SSS at the 2002 International Conference on Parallel Architectures and Compilation Techniques in September 2002.