T. Anthony Marsland

Multi-cut Pruning in Alpha-Beta Search

The efficiency of the αβ-algorithm as a minimax search procedure can be attributed to its effective pruning at so called cut-nodes; ideally only one move is examined there to establish the minimax value. This paper explores the benefits of investing additional search effort at cut-nodes by expanding other move alternatives as well. Our results show a strong correlation between the number of promising move alternatives at cut-nodes and a new principal variation emerging. Furthermore, a new forward pruning method is introduced that uses this additional information to ignore potentially futile subtrees. We also provide experimental results with the new pruning method in the domain of chess.

The Bratko-Kopec Test Revisited

The twenty-four positions of the Bratko-Kopec test (Kopec and Bratko 1982) represent one of several attempts to quantify the playing strength of chess computers and human subjects. Although one may disagree with the choice of test set, question its adequacy and completeness and so on, the fact remains that the designers of computer-chess programs still do not have an acceptable means of estimating the performance of chess programs, without resorting to time-consuming and expensive “matches” against other subjects. Clearly there is considerable scope for improvement, as the success of test sets in related areas like pattern recognition attest.

Shared memory computing on SP2: JIAJIA approach

Distributed shared memory (DSM) is a useful abstraction not only for deploying networks of workstations as a parallel multicomputer but also for increasing the usability of non-uniform memory access multicomputers. It provides an alternative programming model for distributed memory computers. In this paper, we present empirical evaluation of JIAJIA, a software DSM system, on an IBM SP2 cluster. We also discuss the performance of a suite of six widely different applications running under this software and compare them with CVM, another software DSM system. We show that these applications achieve moderate to good speedups and argue that shared memory computing is an attractive alternative to message passing on high performance computer systems, such as SP2 clusters.