Home         Authors   Papers   Year of conference   Themes   Organizations        To MES conference

Synthesis of multiprocessor computing structures based on models of models
Authors
	Kulagin V.P.
	Muraviev N.D.
Date of publication
	2022
DOI
	10.31114/2078-7707-2022-4-21-25
Abstract
	The article demonstrates an approach to modeling multiprocessor computing structures based on network models, and also describes a method for synthesizing new structures based on tensor transformation. Petri nets (PN) are used as network models, as they are intended for modeling parallel processes. As a computational structure, a processor matrix is used, which consists of the simplest processor elements that support the simplest operations of addition and multiplication. The stages of synthesis of computational structures are also demonstrated: decomposition of the original model into linear-basic fragments and a primitive system, calculation of the transformation tensor, definition of synthesis programs, synthesis of a new structure, the external behavior of which does not differ from the original one. A comparative analysis of the synthesized PN-model with the original one was carried out, in which it was shown that the synthesized structure has a more optimal use of computing resources by reducing the idle time of processor elements and their number.
Keywords
	matrix structures, network models, Petri nets, tensor transformation, synthesis.
Library reference
	Kulagin V.P., Muraviev N.D. Synthesis of multiprocessor computing structures based on models of models // Problems of Perspective Micro- and Nanoelectronic Systems Development - 2022. Issue 4. P. 21-25. doi:10.31114/2078-7707-2022-4-21-25
URL of paper
	http://www.mes-conference.ru/data/year2022/pdf/D090.pdf