Parallel Evolution of Parallel Processors - G. Lerman

. .

Résumé :
1. Introduction.- 2. Classification of Parallel Processors.- 2.1. A Brief History of Classification Schemes.- 2.2. The Classification Scheme Used in This Work.- 2.3. A Look at the Classification Characteristics.- 2.4. Information-Gathering Details.- 2.5. An Apology.- 3. Emergent Trends.- 3.1. Applications.- 3.2. Mode of Control.- 3.3. Data Exchange and Synchronization.- 3.4. The Number and Type of PEs.- 3.5. Interconnection Network.- 3.6. Memory Organization.- 3.7. Type of Constructing Institution.- 3.8. Period of Construction.- 3.9. Summary of the Correlations.- 4. Popular Machine Models.- 4.1. Exposing the Complex Patterns.- 4.2. General-Purpose Machines.- 4.3. Model IV - Image (and Signal) Processing SIMD Machines.- 4.4. Model V - Database MIMD Machines, Two Variants.- 4.5. Trends in Commercialization.- 5. The Shape of Things to Come?.- 5.1. Underlying Assumptions.- 5.2. Applications.- 5.3. Control.- 5.4. Data Exchange and Synchronization.- 5.5. Number and Type of PEs.- 5.6. Interconnection Networks.- 5.7. Memory Organization.- 5.8. Sources.- 5.9. Classification of Parallel Computers.- 5.10. Summary.- Appendix: Information about the Systems....

Sommaire:
Study the past, if you would divine the future. -CONFUCIUS A well written, organized, and concise survey is an important tool in any newly emerging field of study. This present text is the first of a new series that has been established to promote the publications of such survey books. A survey serves several needs. Virtually every new research area has its roots in several diverse areas and many of the initial fundamental results are dispersed across a wide range of journals, books, and conferences in many dif? ferent sub fields. A good survey should bring together these results. But just a collection of articles is not enough. Since terminology and notation take many years to become standardized, it is often difficult to master the early papers. In addition, when a new research field has its foundations outside of computer science, all the papers may be difficult to read. Each field has its own view of el? egance and its own method of presenting results. A good survey overcomes such difficulties by presenting results in a notation and terminology that is familiar to most computer scientists. A good survey can give a feel for the whole field. It helps identify trends, both successful and unsuccessful, and it should point new researchers in the right direction....