Recurrence semigroups and their relation to data storage in fast recurrence solvers on parallel machines
There is a wealth of algorithms for solving recurrence systems on parallel machines. However, none of them takes into account the time and cost necessary to supply data to processors and data organization in parallel memory modules. It is shown how algorithms based on semigroups of recurrences lead naturally to the solution of this problem. Furthermore, it is shown that a parallel processor with a shift unit for rearranging data in memories is sufficient for solving recurrence systems.
KeywordsParallel Processor Functional Composition Recurrence System Storage Scheme Arithmetic Element
Unable to display preview. Download preview PDF.
- [Gajs78a]D. D. Gajski, "Processor Arrays for Computing Linear Recurrence Systems," Proc. of the 1978 Int'1. Conf. on Parallel Processing, Bellaire, Michigan, pp. 246–256, 1978.Google Scholar
- [Gajs78b]D. D. Gajski, "An Algorithm for Solving Linear Recurrence Systems on Parallel and Pipelined Machines," University of Illinois, Department of Computer Science Report No. 78–953, Dec. 1978.Google Scholar