Two design patterns for data-parallel computation based on master-slave model

This paper presents two design patterns useful for parallel computations of master-slave model. These patterns are concerned with task management and parallel and distributed data structures. They can be used to help addressi ng the issues of data partition and mapping, dynamic task allocation and ma nagement in parallel programming with the benefit of less programming effor ts and better program structures. The patterns are described in object-orie nted notation, accompanied with illustrative examples in C++. We also provi de our experience in applying these patterns to two scientific simulation p rograms simulating Ising model and plasma physics respectively. Since maste r-slave model is a widely used parallel programming paradigm, the design pa tterns presented in this paper have large potential application in parallel computations. (C) 1999 Elsevier Science B.V. All rights reserved.