On topological scalability of computing systems

Most results on parallel task and system scaling met in the literature are guided by the specific technological and topological implementation. Unique properties of a problem originated from the class of the task solved, from technical, technological and topological implementation of the system, firstly, prevent using these results directly in the analysis and synthesis of the other parallel systems and tasks, and, secondly, hide the general pattern of mutual conditionality of the given parameters of system with demanded performance metrics. On the basis of the model suggested in this paper we try to fill the existing gap with the main goal to assess the impact of the topology of a parallel system on scalability of a system or a task.