Evolutionary design of OAB and AAB communication schedules for interconnection networks

Abstract

Since chip multiprocessors are quickly penetrating new application areas in network and media processing, their interconnection architectures become a subject of sophisticated optimization. One-to-All Broadcast (OAB) and All-to-All Broadcast (AAB) [2] group communications are frequently used in many parallel algorithms and if their overhead cost is excessive, performance degrades rapidly with a processor count. This paper deals with the design of a new application-specific standard genetic algorithm (SGA) and the use of Hybrid parallel Genetic Simulated Annealing (HGSA) to design optimal communication algorithms for an arbitrary topology of the interconnection network. Each of these algorithms is targeted for a different switching technique. The OAB and AAB communication schedules were designed mainly for an asymmetrical AMP [15] network and for the benchmark hypercube network [16] using Store-and-Forward (SF) and Wormhole (WH) switching. © Springer-Verlag Berlin Heidelberg 2006.