Part 2
Regions of Computer Space
Part 2 continues the development of the computer space by delving into details in eight regions of the computer space. Each section opens with a tabulation and discussion of the major subdimensions. The subdimension values are illustrated by actual computer systems, many of which are described in this book. The sections conclude with a series of chapters meant to illustrate various values of the subdimensions and how they correlate with values for other computer space dimensions in actual machines.
The chapters in Part 2 have been selected for their primary emphasis on a single region of computer space. Subsequently, Parts 3 and 4 will examine complete computer systems and treat all computer space dimensions equally.
Section 1 discusses the current major computer implementation technique: microprogramming. While the computer space region should properly be labeled "Implementation Techniques," microprogramming's popularity and richness deserve a separate treatment.
Section 2 examines the region of memory hierarchies and support of multiple processes. The concern for effective utilization of memory has impacted even the smallest microcomputers.
Concurrency to achieve high performance in single-processor systems is the subject of Sec. 3.
The advent of low-cost microcomputers has ignited substantial interest in multiple-processor systems. Section 4 discusses the various ways multiple processors can be interconnected and the important parameters for evaluating the effectiveness of the PMS structure. Three major categories of multiple-processor structures are identified and illustrated: tightly coupled multiprocessors communicating via address space; loosely coupled distributed multiprocessors communicating via messages, all Pc's working on one task; and networks communicating via messages, each Pc working on different tasks. Section 4 gives examples of the first two types of multiple-processor systems.
Network technology has advanced so significantly in the last decade that one section, Sec. 5, is devoted solely to the network region of the computer space.
The concern for reliable computing has been with us from the earliest days. The need for reliable computers continues as our dependency on computers grows. Now all but the smallest computers have introduced redundancy to improve system reliability and/or maintainability. The fault-tolerant region of computer space in Sec. 6 is one of the least well-formed of any treated in the book. However, this will change as more fault-tolerant systems are built and experience accumulates.
The final two sections of Part 2 discuss related regions of computer space. Section 7 examines computers intended to execute a single higher-level programming language. Since the programming environment is completely specified (in contrast with the open-ended environments found in general-purpose systems), design decisions can be made to favor specialization. Section 8 looks at another constrained environment, that of personal computers. Personal computing systems are dedicated to providing a rich, responsive programming environment to a single user.
149