Excerpt from Memory Management in Symunix II: A Design for Large-Scale Shared Memory Multiprocessors The remainder of the paper is organized as follows. Section 2 outlines the parallel pro cessing requirements that most influenced the design. Section 3 describes the user's memory model and the system interface. Some aspects of the internal design, including performance issues, are presented in Section 4. The concluding Section 5 gives a status report. About the Publisher Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.

In general, distributed systems can be classified into Distributed File Systems (DFS) and Distributed Operating Systems (DOS). The survey which follows distinguishes be tween DFS approaches in Chapters 2-3, and DOS approaches in Chapters 4-5. Within DFS and DOS, I further distinguish "traditional" and object-oriented approaches. A traditional approach is one where processes are the active components in the systems and where the name space is hierarchically organized. In a centralized environment, UNIX would be a good example of a traditional approach. On the other hand, an object-oriented approach deals with objects in which all information is encapsulated. Some systems of importance do not fit into the DFS/DOS classification. I call these systems "closely related" and put them into Chapter 6. Chapter 7 contains a table of comparison. This table gives a lucid overview summarizing the information provided and allowing for quick access. The last chapter is added for the sake of completeness. It contains very brief descriptions of other related systems. These systems are of minor interest or do not provide transparency at all. Sometimes I had to assign a system to this chapter simply for lack of adequate information about it.

Parallel processing is seen today as the means to improve the power of computing facilities by breaking the Von Neumann bottleneck of conventional sequential computer architectures. By defining appropriate parallel computation models definite advantages can be obtained. Parallel processing is the center of the research in Europe in the field of Information Processing Systems so the CEC has funded the ESPRIT Supemode project to develop a low cost, high performance, multiprocessor machine. The result of this project is a modular, reconfigurable architecture based on !NMOS transputers: T.Node. This machine can be considered as a research, industrial and commercial success. The CEC has decided to continue to encourage manufacturers as well as research and end-users of transputers by funding other projects in this field. This book presents course papers of the Eurocourse given at the Joint Research Centre in ISPRA (Italy) from the 4th to 8 of November 1991. First we present an overview of various trends in the design of parallel architectures and specially of the T.Node with it's software development environments, new distributed system aspects and also new hardware extensions based on the !NMOS T9000 processor. In a second part, we review some real case applications in the field of image synthesis, image processing, signal processing, terrain modeling, particle physics simulation and also enhanced parallel and distributed numerical methods on T.Node.