Design Choices/Engineering Trade-Offs

T his special issue addresses the business of producing computing machines and the choices computer architects and implementors must make to succeed in the marketplace. This focus represents the sober reflection that computer architecture is not yet a science, even though parts of the discipline can be considered science and most of it often involves scientific methodology. To a large extent, the field is not driven by an interest in finding order in what appears to be chaos, but rather by an interest in producing a product that people will buy. To produce that product, architects and implementors have a budget, whether measured in dollars, gates, person-years, or similar entities, with which to achieve a certain functional capability at a specified performance level. Within those constraints, many choices are available. This issue of Computer examines these choices within the context of specific machine designs. Some of these choices and a framework for their consideration are identified below.