THE METHOD OF PARALLELIZATION OF DYNAMIC SEQUENTIAL SYSTEM PROGRAMS USING NETWORKS OF MULTIFACETED PROCESSES

Abstract

The article examines the peculiarities of dynamic sequential system programs and identifies the main problems of their effective parallelization. It is shown that the complexity of automatic parallelization of such programs is due to the presence of complex dependencies between data, irregular structure of calculations, dynamic creation of tasks, as well as the need to synchronize access to shared resources. These factors significantly complicate the use of traditional approaches to parallel execution and require the development of specialized methods for organizing calculations.

The paper proposes a method of parallelizing dynamic sequential system programs based on the use of networks of multifaceted processes. The proposed approach makes it possible to formalize the structure of calculations, decompose the program into independent or partially independent fragments, and organize their execution in the form of a system of interacting processes that can be executed in parallel. This ensures more efficient use of computing resources and increases the level of scalability of applications.

A model for representing a sequential system program in the form of a network of interacting processes, within which the mechanisms of their interaction and synchronization are defined, has also been developed. The proposed mechanisms ensure correct data exchange between processes, consistency of execution and prevention of access conflicts to shared resources. The implementation of such a model allows more efficient organization of parallel execution of programs in multi-core computing systems.

A study of the effectiveness of the proposed method was conducted, the results of which confirm the expediency of using networks of multifaceted processes to increase the productivity of executing dynamic system programs. The use of the proposed approach contributes to the optimization of computing processes and more rational use of hardware resources of modern multicore systems.

A promising direction of further research is the study of the possibilities of applying the proposed approach for parallelization of processes in distributed computing environments.