Developing parallel procedures for line simplification

This paper explores parallel programming issues that are relevant to the efficient implementation of spatial data handling procedures on current parallel computers through sample implementations of the Douglas line simplification procedure. Using source code-equivalent implementations of the Douglas procedure, this paper analyses the performance characteristics of two parallel implementations, compares their performance characteristics to those of a sequential implementation, and identifies critical components of the parallel implementations that enhance or inhibit their overall performance values. The results of this work show that the selection of appropriate interprocessor communication and load balancing strategies are crucial to obtaining large speedup values over comparable sequential implementations.