Work-load balancing in highly parallel depth-first search

Abstract

Among the various approaches for parallel depth-first search (DFS), the stack-splitting schemes are most popular. However, as shown in the paper, dynamical stack-splitting is not suitable for massively parallel systems with several hundred processors. Initial work-load imbalances and work packets of dissimilar sizes cause a high communication overhead. We compare work-load balancing strategies of two depth-first searches and propose a scheme that uses fine-grained fixed-sized work packets. In its iterative-deepening variant (named AIDA*) the global workload distribution improves from one iteration to the next. As a consequence, the communication overhead decreases with increasing search time.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>