Title: Parallel Processing Implications for Software Engineering
Abstract: Abstract This article considers parallel systems to be a broader topic than the mapping of single algorithms to parallel processors. This strategy allows for a discussion of general principles, rather than the techniques that apply to specific machines. Much of the present parallel hardware industry provides machines that do not support a parallel machine model as needed for software engineering practices, therefore, this strategy is necessary. Massively parallel machines are organized for selected algorithm approaches that attain high performance at the sacrifice of software productivity because unacceptable application constraints result when building complete parallel systems with multiple algorithms and application domains. For these machines the goals of software engineering practices cannot be met. Some parallel machines do support a suitable parallel machine model for software engineering, but these have a maximum processor count which limits peak performance. To better discuss parallel systems two styles are introduced: intensive and extensive . Extensive concurrent applications are those that are too complex and interrelated to have a recognizable and predictable response structure. Intensive algorithms are those that have predictable data and control flow paths. Both styles require software engineering methods and tools and are sufficiently different from each other so that the style must be stated for any rational and reasonable discussion.
Publication Year: 2002
Publication Date: 2002-01-15
Language: en
Type: other
Indexed In: ['crossref']
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot