Fuzzy logic based adaptive MPSoC for balanced energy and throughput

In this paper a Fuzzy logic based Controller has been proposed to reconfigure multi-processor system on chip (MPSoC)architecture according to the workload requirements. Mamdani and Sugeno Inference Engines are compared andanalyzed to enhance the smooth regulation of reconfiguration in design space....

Full description

Saved in:
Bibliographic Details
Main Authors: Adeel Ahmad, Muhammad Yasir Qadri, Nadia N. Qadri
Format: Article
Language:English
Published: Elsevier 2016-08-01
Series:Kuwait Journal of Science
Subjects:
Online Access:https://journalskuwait.org/kjs/index.php/KJS/article/view/924
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper a Fuzzy logic based Controller has been proposed to reconfigure multi-processor system on chip (MPSoC)architecture according to the workload requirements. Mamdani and Sugeno Inference Engines are compared andanalyzed to enhance the smooth regulation of reconfiguration in design space. The proposed MPSoC platform consistsof 16 cores, where each core has private L1 cache and a shared L2 cache. The design space exploration parametersevaluated are CPUs operating frequency, number of cores, throughput and energy consumption. Marssx86 simulatoris used to build the target machine and SPLASH-2 benchmarks are executed on this configured architecture. MamdaniInference Engine for the said implementation has shown up to 14%-23% decrease in Energy-Delay2Product (ED2P) forvarious benchmark applications. Hence the performance of Mamdani inference engine was found to be better than theSugeno one for this particular implementation.
ISSN:2307-4108
2307-4116