A method of optimising data distribution on a loosely coupled multiprocessor system

The maximum speedup of a multiprocessor system is limited by the sequential part of an algorithm and in loosely coupled processor systems, a large part of this sequentiality is caused by the communication between processors. As this communication is dependent on the distribution of data the data distribution must be optimised in order to achieve the maximum speedup. In this paper we present a new method of achieving this optimisation for loosely coupled multiprocessors using a branch and bound technique based on the Moore-Skelboe interval arithmetic. Two key issues of this algorithm have been addressed, namely, the branch selection criterion, and an efficient data structure to keep track of the branches. A blocked data structure is shown to have the desirable properties of dynamic list length, and minimal disk thrashing. When this method is applied to an FFT algorithm running on a pipeline of transputers, the optimised data distribution is shown to yield an improved speedup over the equal and random distributions, which do not take into account communication overheads.<
>