A Poly-Algorithm for Efficient Parallel Matrix Multiplication on Metacomputing Platforms

Abstract

There is today an increasing diversity of parallel execution supports. New characteristics are influencing the execution of parallel applications, like for instance the hierarchical structure and the heterogeneity of the processors. Designing efficient software that takes full advantages of such systems remains difficult Indeed, solving a target problem by using a single algorithm is not always efficient on any computational support. We present in this paper a poly-algorithmic approach for selecting the most suitable algorithm among various ones for given problem size and available metacomputing platform, like hierarchical clusters of SMPs (Symmetric Multi-Processors) which become more and more popular. Our principal objective here is to illustrate such an approach on the well-known matrix multiplication problem which is one of the most important basic numerical kernels