ARTICLE CRITIQUE

Article Critique on Parallel Computing

Article Critique on Parallel Computing

Introduction

This paper presents a critical analysis of an article. The title of the article is “Effective Scheduling in a Mixed Parallel and Sequential Computing Environment”. This article is written by a team of authors, B.B. Zhou, X.Qu, and R.P. Brent. All of the authors are employed at Australian National University. This article was presented at the Sixth Euromicro Workshop in 1998 and was published in the Journal called, Parallel and Distributed Processing. At the end of this article, there are 16 bibliographic entries, numbered as in Chicago-Turabian style. This article is basically written for software engineers and network engineers, who are enrolled at Australian National University, facilitating them gain an insight into what problems they may face and how the issues in parallel computing can be handled effectively (Culler, 2008).

Analysis

The basic area of problem, as mentioned in this article, relates to inefficiency of batch systems in relation to the execution and scheduling of simultaneous parallel tasks. To address this problem, authors have come up with the idea of a two level scheduling scheme, meant to streamline the parallel and sequential processes in such a way that they do not overlap. The authors have followed the traditional and typical pattern of research paper and the aims and objectives are presented in the introductory paragraphs. In this article the authors have described two- level scheduling for mixed and sequential computing machines. In this article, the authors have discussed about two principles that govern the effective performance of computing machines - one is the coordination between scheduling of parallel programmes and the performance of this parallel system (Culler, 2008).

Over the last thirty years, parallel computing has evolved very fast, too fast to follow the latest technological developments: just out in the market, the machines are obsolete, so do not talk about books that talk about the machines. Nevertheless, the concepts remain, and we want to discover the utility of scheduling in parallel systems. The parallel algorithm borrows heavily from the traditional algorithmic problems in its design, analysis, and study of complexity. The results are sometimes less accurate because the problems are more recent, and also more difficult. Basically, there is still a new dimension, an additional degree of freedom with the simultaneous use of multiple resources. The authors have identified core fundamental knowledge, largely independent of architectural detail. They have described models, algorithms and techniques (scheduling, compilation) that will not be obsolete in three years (a promise), not in ten years (it is a profession of faith). Parallel computing has seen, we said, a rapid evolution. It should be added chaotic, with highs. It turns out that we are currently top of the wave: the parallelism is ubiquitous, and this at both ends of the spectrum. At the microscopic level, processors multiply pipelined arithmetic units on a single integrated circuit. In recent years CPUs have hardware limitations that necessitate a special cooling. At this point the race to shrink the silicon and increase the ...