Eecient Multicast on Wormhole Switch-based Irregular Networks of Workstations and Processor Clusters

High bandwidth and low latency switches are commercially available. Using these switches, it has become a popular and cost-eeective way to interconnect workstations and processor clusters together as a parallel computing platform. The interconnection topology is usually irregular. On such systems, multicast is an important collective operation. Two steps are involved in a multicast: (1) the source node sends the multicast message to the destinations which are connected to a switch directly or are the leader of a processor cluster, and (2) the leader node of each cluster sends the message to other destinations in the same cluster. In this paper, we propose two unicast-based multicast algorithms. Algorithm Multicast 1 performs those two steps sequentially; while Algorithm Multicast 2 overlaps them. Performance of the two algorithms will be evaluated and compared. It is found that Algorithm Multicast 2 performs better when the number of participating processor clusters is small and the number of destinations of a multicast is large.