Prototyping High-Performance Parallel Computing Applications in Proteus

This paper explores the use of Proteus, an architecture-independent language suitable for prototyping time-sensitive parallel and distributed programs. Proteus is a high-level imperative notation based on sets and sequences with succinct yet powerful constructs for the parallel composition of processes communicating through shared memory. Several di erent parallel algorithms for N-body simulation in molecular dynamics are presented in Proteus, illustrating how Proteus provides a common foundation for expressing the various parallel programming models. This common foundation supports the construction of high-performance computing applications across a wide range of parallel machines through a development methodology in which prototype parallel programs can be tested and evolved without the use of machine-speci c languages. To transform prototypes to implementations on speci c architectures, program re nement techniques are utilized. Re nement strategies are illustrated that target broad-spectrum parallel intermediate languages, and their viability is demonstrated by re ning an N-body algorithm to data-parallel intermediate code. Time-sensitive variants of a parallel N-body algorithm are also described to illustrate how Proteus allows the expression of resource requirements through real-time constraints as well as progress constraints which abstractly specify the distribution of computational resources.