Highly fault-tolerant hypercube multicomputer

Abstract

A strongly fault-tolerant design for a d-dimentrun aucountsional hypercube multiprocessor is presented and its reconfigurability examined. The augmented hypercube has a spare node connected to each node of a subcube of dimension i, and the spare nodes are also connected as a (d - i)-dimensional hypercube. By utilising the circuit-switched capabilities of the communication modules of the spare nodes, a large number of faulty nodes and faulty links can be tolerated. Both theoretical and experimental results are presented. Compared with other proposed schemes the approach can tolerate significantly more faulty nodes and faulty links with a low overhead and no performance degradation.