Knowledgeable network-on-chip accelerator for fast and accurate simulations using supervised learning algorithms and multiprocessing
by Anil Kumar; Basavaraj Talawar
International Journal of Intelligent Engineering Informatics (IJIEI), Vol. 10, No. 2, 2022

Abstract: In a multi-processor system-on-chip (MPSoC) environment, networks-on- chip (NoC) is becoming the de-facto scaling communication technique. One of the most significant techniques used in NoC for analysing and testing new architectures is simulations. Simulation is one of the main tools used in NoC for analysing and testing new architectures. To achieve the best performance vs. cost tradeoff, simulators have become an essential tool. Software simulators are too slow for evaluating large-scale NoCs. To overcome this problem we propose an NoC Accelerator named knowledgeable network-on-chip accelerator (KNoC) which can be used to analyse various NoC architectures. The proposed accelerator is built using machine learning (ML) algorithms and multiprocessing to predict the design parameters of NoCs with a fixed and accurate delay between nodes of large-scale architectures. The KNoC results were compared to the widely used cycle-accurate Booksim simulator. KNoC showed an error rate of less than 6% and an overall speedup of up to 12 Kx.

Online publication date: Fri, 30-Sep-2022

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Intelligent Engineering Informatics (IJIEI):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com