A CASE STUDY ON CLUSTER BASED POWER-AWARE MAPPING STRATEGY FOR 2D NoC
Abstract
vioft2nntf2t|tblJournal|Abstract_paper|0xf4fffd7c22000000058f050001000400
Network on Chip (NoC) is a growing and prominent paradigm which improves the power and performance of the System on Chip (SoC). Application mapping is one of the major challenges in NoC which maps the various Intellectual Property (IP) cores to standard network topology. Among the various application mapping methods, Integer Linear programming (ILP) is one of the static mapping methods, which finds optimum communication cost. However, it consumes longer computation time. To overcome this limitation, cluster based mapping using KL algorithm has been introduced and it acts poorly at partitioning cut degree. Based on these studies, we propose Fidducia-Mattheyses (FM) algorithm for multi clustering to optimize power consumption and communication cost for different benchmarks of NoC. The effectiveness of the proposed method is verified through VOPD, MPEG 4 and PIP benchmarks. Experimental results show a 4.4% and 34% improvement on communication cost and power consumption respectively for FM algorithm with MPEG 4. However, for VOPD the communication cost and total power consumption is improved with 27% and 35% respectively. On the other hand, PIP benchmark offers identical results in total power consumed and communication cost minimization with existing methods.

Authors
S Salini1, A Aravindhan2, G Lakshminarayanan3
Saintgits College of Engineering, India1, Saintgits College of Engineering, India2, National Institute of Technology, Tiruchirappalli, India

Keywords
Application Mapping, Integer Linear Programming (ILP), Cluster Based Mapping, Fidducia-Mattheyses (FM) Algorithm
Yearly Full Views
JanuaryFebruaryMarchAprilMayJuneJulyAugustSeptemberOctoberNovemberDecember
000000000000
Published By :
ICTACT
Published In :
ICTACT Journal on Microelectronics
( Volume: 2 , Issue: 4 , Pages: 315-322 )
Date of Publication :
January 2017
Page Views :
159
Full Text Views :

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.