Journal
Scientific and technical journal of information technologies, mechanics and optics
UDK519.687+004.75
Issue:2 (66)
We develop a performance model of solving computational tasks with variable size on distributed systems. For this purpose we extend a performance model of systems with schedule to allow for stochastic nature of global optimization problem. In addition, we consider an experimental method to estimate task size and calculate a confidence interval for efficiency. Finally, we show that a real distributed system solving a cluster conformation problem follows the model within estimated accuracy.