Abstract
The article discusses the issues of classification selection of preferred parallelization algorithms (with minimal execution time) implemented in parallel software development tools for multi-core (multiprocessor) computing systems with shared memory, based on the collected training statistical information of the parameters of the execution of cycles, previous program launches. The classification (decisive) rule for selecting the preferred algorithm in the learning process can be built only based on information from the training sample of previous runs, this information should not just be remembered, but should be generalized and transformed into an image most similar to the new recognizable image characterizing an unknown program cycle. To this end, the article presents three methods that allow, based on the mathematical apparatus of probability theory, to sequentially collect and transform training samples for each cyclic section of the program to the form of an image providing a functional relationship between the number of iterations of the current cycle and the preferred parallelization algorithm. The obtained dependence allows automatic selection (in real time) of the preferred parallelization algorithm based on classification selection by the input parameter (number of iterations) of the cycle in the generated project profile consisting of a set of tuples of preferred parallelization algorithms. The purpose of this article is to assess the impact of the execution time of parallelized cyclic sections of the target program, using the proposed method of automated selection of preferred algorithms, with multithreaded parallel execution of the program in multi-core (multiprocessor) PCs on the results of simulation of combat operations
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Journal of Siberian Federal University. Engineering & Technologies
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
