I.c. Legrand, A. Gellrich, U. Gensch, H. Leich, P. Wegner

Introduction The high event and data rates of future high energy physics experiments require complex on-line parallel processing systems with large I/O bandwidth capabilities. The design of such dedicated distributed processing systems requires detailed simulations which must be able to describe many processing nodes running complex real-time data processing tasks and which are interconnected by a large bandwidth communication network. The task of executing simulations requires, as a first step, the design of a model able to capture the dynamic behavior of the system. We are studying asynchronous distributed data processing systems for which the real processing time is very much input data dependent. By using a multi-thread approach for modelling such systems, which is both close to reality and well suited to the object oriented paradigm, the structure of the simulation program provides a convenient way to test different hardware architectures and software protocols