%T A tool for optimisation of program execution in dynamic topology systems %A Tomasz Kalinowski %E Andr\[`e] W. P. Bakkers %B Proceedings of WoTUG\-20: Parallel Programming and Java %X In this paper, we present a tool for optimisation of execution of parallel programs in distributed memory multi\-processor systems with dynamic interconnection networks. The programs are described as Directed Acyclic Graphs (DAGs). The tool allows to compare simulated execution times for different task scheduling heuristics, target system topologies and communication models. A list scheduling algorithm, which has been applied, accounts for dynamic changes of interconnection structure. We demonstrate the efficiency of dynamic networks by comparing schedules obtained for dynamic and fixed topology systems. We propose a method of validating simulation results in a target system composed of T9000 transputers. The method relies on comparison of simulation results with execution times of synthetic OCCAM applications in the target system. The comparison indicates that assumptions taken on program execution and system model hold in the system under investigation.