Automatic generation of GSPN models for the performance evaluation of heterogeneous applications.
Cosimo Anglano · 1997
The allocation of the tasks of heterogeneous applications on the machines of the system is crucial to achieve adequate performance. The development of effective allocations is dependent upon the possibility of obtaining accurate predictions of the performance of the application. In this paper we present a technique for the automatic generation of performance models of heterogeneous applications starting from Zoom, a high-level graphical formalism for the representation of heterogeneous applications. Generalized Stochastic Petri Nets are used to specify these performance models. A testbed heterogeneous application is used to illustrate the proposed technique. 1. Introduction In the last decade distributed heterogeneous systems, consisting of a collection of autonomous high-performance machines connected by a fast communication network, have emerged as a powerful and cost-effective platform for the execution of computationally intensive applications. The feasibility of the heterogeneous...