Efficient Broadcast in All-Port Wormhole-Routed Hypercubes

A method to reduce broadcast time in wormholerouted hypercube systems is described. The method takes advantage of the distance insensitivity of wormhole routing and the presence of multiple ports between processors and their routers. Performance results from an nCUBE-2 multicomputer are given that demonstrate the advantage of the method over the traditional spanning binomial tree approach. 1 Introduction Massively parallel computers (MPCs) are characterized by the distribution of memory among an ensemble of nodes, whose processors communicate through a network. MPCs are scalable because, as the number of nodes in the system increases, the total communication bandwidth, memory bandwidth, and processing capability of the system also increase. Efficient communication among nodes is critical to the performance of MPCs. Communication may be point-to-point or collective, depending on whether exactly two or more than two processes participate. Examples of collective communication include da..