Chapter 1. Architecture and Implementation

Computer science often distinguishes between abstraction and implementation i.e., between the general and the particular. We may examine any computer system at two major levels: its architecture and its organization. Although numerous books convey both of these levels in their titles and contents, we are going to concentrate on architecture in this book. We first direct our readers toward an understanding of the distinction between these levels.

In the first decades of the history of computers, the sporadic emergence of new ideas and new companies resulted in a jumbled succession of disparate approaches to computer design. The design of the IBM^® System/360^™ series by Amdahl and his team, however, marked not only the trend-setting idea of a family line of computers but also a clear articulation of architecture as distinct from implementation:

• The architecture of a computer system is the abstraction equivalent to the user-visible interface: the structure and the operation of the system as viewed by the assembly language programmer and the compiler-writer. If an architecture is well-designed, well-engineered to adapt to future technologies, and well-liked by the market, it may persist for a decade or longer.

• An implementation is the realization and construction of that interface and structure out of specific hardware (and possibly software) components. Because of technological advances, any particular implementation (i.e., one model) may only be actively marketed for a relatively short period of time.

Several different implementations of an architecture may appear over a period of years. Each can offer different trade-offs among cost, performance, and convenience, but all will present the same interface to the assembly language programmer. Such consistency over time, despite technological change, has clearly helped computer system manufacturers to develop brand loyalty and facilitate the development of software as an allied industry.