Parallel Randomized Best-First Search

Abstract

We describe a novel parallel randomized search algorithm for twoplayer games. The algorithm is a randomized version of Korf and Chickering’s best-first search. Randomization both fixes a defect in the original algorithm and introduces significant parallelism. An experimental evaluation demonstrates that the algorithm is efficient (in terms of the number of search-tree vertices that it visits) and highly parallel. On incremental random game trees the algorithm outperforms Alpha-Beta, and speeds up by a factor of 18 (using 35 processors). In comparison, Jamboree [Kuszmaul ’95], speeds up by only a factor of 6. We have also evaluated the algorithm in a Chess-playing program using the boardevaluation code from Crafty, an existing Alpha-Beta-based program. On a single processor our program is slower than Crafty; with multiple processors it outper-

3 Figures and Tables

Cite this paper

@inproceedings{Shoham2001ParallelRB, title={Parallel Randomized Best-First Search}, author={Yaron Shoham and Sivan Toledo}, year={2001} }