P-CAM: a framework for parallel complex systems simulations

History has taught that the design and implementation of an efficient paral lel simulation program is a tedious and error prone process. Methods that c an circumvent the parallelization steps in this process are usually warmly welcomed by parallel simulation architects. In this paper we introduce a Pa rallel Cellular Automata Modeling, environment (P-CAM) for doing spatial lo ad balancing on arbitrary connected grids or task graphs. This environment adopts an object oriented application framework in which we can instantiate a variety of simulation problems. We have implemented a kernel, based on t his framework, which facilitates dynamic load balancing and supports proces s migration and irregular interprocess communication patterns. The design o f the kernel enables a transparent implementation of complex systems models onto arbitrary parallel computer systems. We show that the design of a par allel simulation program can be assisted by using the P-CAM kernel. (C) 199 9 Elsevier Science B.V. All rights reserved.