A method for efficient multiplication of dense matrix using an approach of parallel and recursive computing

摘要

In this paper we describe a computational model oriented to the efficient processing of dense matrix using an approach parallel and linear recursive synchronous computing. The architecture, design, and performance of this conceptual model are described, and it is denominated Parallel and Recursive Machine or PRM. This model has been used to deal the problem of functions list evaluation. The model is based on important computation notions as the parallelism and the recursive theory, which both govern its organization and operation mode. The parallelism in fine grain is exploited by means synchronous processing across of pipelines of processors, the concurrency and the execution control by means of dataflow. The structure of the PRM is organized by means a stack of recursive levels that control the processing massive of dense matrix as a large list of linear recursive functions.