Communications of the ACM - Special section on computer architecture
Data-flow algorithms for parallel matrix computation
Communications of the ACM
Computer
Hi-index | 14.98 |
In this paper we extend a model of Karp and Miller for parallel computation. We show that the extended model is deterministic, in the sense that under different scheduling regimes each process in the computation consumes the same input and generates the same output. Moreover, if the computation halts, the final state is independent of scheduling. The model is applied to the generation of precedence graphs, from which lower time bounds may be deduced, and to the synchronization of systolic arrays by local rather than global control.