Please use this identifier to cite or link to this item: http://localhost:8080/xmlui/handle/123456789/13808
Title: Performance Evaluation of Networks On-chip Topologies
Authors: Ziraoui, Zakaria
Keywords: Performance evaluation
Network-On-Chip
Mesh
Express Cube
Simulation
Discrete-Event Simulation
Issue Date: 2021
Publisher: Université Blida 1
Abstract: The main objective of this project is to conduct a performance evaluation of some well-known topologies that have been proposed for Networks-on-Chip (NoC) including the Mesh and Express Cube along with its variations. This later is based on a simple Mesh interconnection network augmented by long-hop wires called express channels to enable bypassing of intermediate hops for non-local communication. Whereas existing studies have focused on the graph-theoretical merits of such topologies, our present study examines the performance of networks-on-chip taking into account the constraints imposed by implementation technology. The most relevant constraint for NoC systems is the wiring density of the chip. To achieve our goal, we have developed a simulation model using the discrete-event simulation technique. Extensive simulation experiments have been performed and the collected results have then been analysed using statistical methods. Our results reveal that while the Express Cube has superior performance when technological constraints are ignored due to its richer connectivity, its performance degrades considerably compared to the Mesh when technological constraints are taken into consideration. Our results also indicate that the 4 Hops variation of the Express Cube is the more suitable for Networks-On-Chip among the Express Cube variations as it is a better fit for designs with large number of cores. Keywords: Performance evaluation, Network-On-Chip, Mesh, Express Cube, Simulation, Discrete-Event Simulation.
Description: ill., Bibliogr.
URI: http://di.univ-blida.dz:8080/jspui/handle/123456789/13808
Appears in Collections:Mémoires de Master

Files in This Item:
File Description SizeFormat 
Ziraoui Zakaria.pdf2,2 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.