Simulation of Multithreaded Algorithms Using Petri-Object Models

Multithreaded programming used for the development of faster algorithms is a very effective method. However, the designing, testing and debugging of nontrivial programs are not easy and need to be improved. Stochastic behavior of threads entails their conflicts and in some cases the unpredictable result of the program. Stochastic Petri nets are widely used for the investigation of concurrent processes in many areas: manufacturing, computer systems, workflow management. In this research stochastic multichannel Petri net is considered as a tool for multithreaded programs modeling. The correspondence between main instructions of multithreaded program and fragments of stochastic Petri net is discovered. Petri-object model’s formalization and software are used for complicated models’ constructions. This approach allows duplicating objects with the same dynamics and aggregating them in model. Models that present the concurrent functioning of multithreaded Java programs are considered. Model’s verification indicates its accuracy. The results of experimental research of these models show a strong impact the values of time delay.