Abstract
This chapter discusses some of the similarities and differences among the many different interconnection strategies used in networks and arrays. It also reviews the relatively few evaluations and comparisons that have been made. A variety of criteria have been suggested and used—empirical, structural, and formal—to construct and evaluate multicomputers. The chapter presents a comparison of networks on a variety of formal, empirical, and structural grounds. Networks are sometimes compared by using formal characteristics—the longest shortest path between nodes, average distance between nodes, number of nodes, number of links, and number of links per node. But a well-written program will be just that program that puts processes that must communicate frequently as close together as possible. The actual performance of networks cannot be compared until meaningful problems and mixes of programs are run through them in empirical tests. They are too complex to be analyzed and too large to be simulated. There is a number of striking structural similarities among networks that on the surface appear quite different.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call