Analytical modeling of parallel applications in heterogeneous computing environments: A study of cholesky factorization

Abstract

Achieving satisfactory performance results in heterogeneous computing environments requires a careful workload assignment. The use of approximate analytical models can help to understand which are the parameters that mostly affect performance. In this paper we will show how to study analytically the behavior of a Cholesky factorization code running in a heterogeneous NOW under the PVM run-time system. Firstly the Cholesky factorization algorithm is introduced, and an analysis of the load distribution is performed. Then the construction of the analytic model of the application is described. Finally, the obtained results are compared to the performance figures obtained by executing the program in the real computing environment.