direct stochastic modeling

When designing a parallel computing system for completing a given set of interactions, the robot is an important characteristic of social security. As well as the basic indicators of the higher number of computing systems, to protect yourself from technical problems, the main indicator of the higher value of the system is to assess the accuracy of the parallel calculation system of specific tasks. Tsya otsіnka seamlessly due to the effectiveness of vikoristannya obzumyudlyavnyh systems.

To solve the forecasting problem, a software package has been developed in full, which is based on mathematical models, methods and algorithms of direct stochastic modeling and tiered stochastic modeling, which are used to estimate the execution time of folding software systems in multiprocessor computer systems. The given software package calculates the average value and the distribution function of the execution time of a set of interrelated tasks on homogeneous resources of a parallel computing system.

To improve the accuracy of the estimation of the time of execution of complex software complexes on parallel computers, an algorithm for the uniform distribution of vertices of a graph of a given set of interrelated works is developed. This algorithm is used in the method of plywood stochastic simulation of multiprocessor computer systems.