Empirical Performance Analysis of Parallel Programs on Wired and Wireless Local Area Networks
Empirical Performance Analysis of Parallel Programs on Wired and Wireless Local Area Networks
No Thumbnail Available
Date
2014-05
Authors
Saidu, Isah Charles
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
High speed and reduced cost of hardware have always been a key motivating factor for innovative concepts in computer performance. At the moment there is a shift from the way computers are built which involves trying to make a single processor run faster to combining several numbers of processors/processing units to simultaneously solve a task. A special kind of High performance computer known as the Beowulf cluster is the result of this innovation. This thesis focuses on measuring performance of parallel programs deployed on wired and wireless networks. Our experiments were conducted on Beowulf clusters. It has been shown empirically in this work that parallel programs written for Beowulf clusters on Wireless Local Area Network (LAN: IEEE 802.11) do not gain appreciable speedup as the number of processing nodes increases compared to the same parallel programs written for the same Beowulf clusters but on wired LAN. In order to improve performance of parallel programs on wireless networks without exploiting any thread level parallelism, a multiprogramming approach was used to localize communication and reduce network overhead thereby improving performance
Description
A Thesis submitted to the School of Postgraduate Studies, Ahmadu Bello University, Zaria In Partial fulfilment of the requirements for the Award of M.Sc Degree in Computer Science. Department of Mathematics, Faculty of Science, Ahmadu Bello University, Zaria May, 2014
Keywords
Empirical,, Performance,, Analysis,, Parallel Programs,, Wired,, Wireless Local Area Networks