Authors
Veselov A.
|
Degree
|
Dr of Technique, Department of Electronic Computer, Tver State Technical University |
|---|---|
|
E-mail
|
evm.tstu@gmail.com |
|
Location
|
Tver |
|
Articles
|
Experimental studies of a distributed model based on Petri nets
The paper presents the results of experimental studies of the effectiveness of the application of distributed
models in comparison with their monolithic analogs. As a basic model, models based on the
mathematical apparatus of Petri nets are used. In particular, special expansion of ordinary, temporary
Petri nets with inhibitory arcs, oriented to modeling of digital devices of automation and computer facilities.
A description of the indicators used to evaluate the effectiveness of monolithic and distributed
models is given. The advantages and disadvantages of these types of models are discussed. The short
description of this model, indexes for assessment of effectiveness and conditions of carrying out an
experiment is given. Analysis of effectiveness of the distributed models allowed to select two feature.
The first is that, in process of increase in their size, the efficiency factor of the distributed models continuously
increases. At the same time, models with the smaller number of hierarchical levels are more
efficient. The second feature is that, in process of further increase in the sizes of the distributed models,
the effectiveness on speed of models with a large number of hierarchical levels becomes not only
equal, but to exceed effectiveness of models with their smaller quantity.
It is shown, that high-speed performance of the distributed models can be two orders more, and time
of their loading is tens times less, than similar indices at monolithic models.
Read more...
Influence of additional hierarchical levels on the working characteristics of distributed modelsIn designing modern computer equipment and digital electronics, the use of simulation models is of great importance. At first, monolithic models were widely used for this. However, they worked well only when their size was relatively small. Because of it developers began to refuse gradually use of monolithic models and to pass to use of the distributed models allowing to increase their speed and to expand borders of their admissible sizes. At the same time, they begin to pay special attention to hierarchical distributed models, which provide the opportunity to investigate the behavior of the created devices at different levels of detail. Similar models made it possible to noticeably expand the permissible boundaries of their sizes and increase the speed of work. However, such distributed models have the disadvantage that their effectiveness is noticeably dependent not only on the number of components included in their composition, but also on the size of these components. he paper presents the results of a study of the effect of introducing an additional upper hierarchical level on the performance of distributed models based on Petri networks. The use of such a method of modifying distributed models leads to an increase in their speed in a wide range of changes in their sizes. At the same time, the most significant effect achieved in distributed models containing a large number of small components. The maximum speed of the thus modified models can be an order of magnitude higher than that of the non-modified ones. As a result, in addition to the overall increase in the efficiency of the modified hierarchical distributed models, this also led to a significant equalization of the performance of the modified distributed models with subordinate components of different sizes. Read more... |
