In the Beginning

Because modules were DEC's first product, and for many years their major product, it is appropriate to study the history of DEC's modules and the influence of technology on their development. The history of modules is a subset of the history of computers, and many of the views of computers expressed in Chapter 1 apply as readily to modules. In particular, the Structural View and the Packaging Levels-of-Integration View plainly apply. Further, a study of module history shows the effects of progress in semiconductor technology, as discussed in Chapter 2, and demonstrates on a small scale many of the packaging and manufacturing concepts discussed in Chapter 3.

With the advent of microprocessors, the distinction between a module and a computer has become blurred, and complete computer systems have become available at the printed circuit board/module level of packaging integration. The structural levels (Chapter 1, Figure 1) found on a single module have changed from solely circuit level to logic level, then to register transfer level, and finally to processor-memory-switch level. These developments will be explored more fully in Part IV, "The Evolution of Computer Building Blocks"; the discussion here is limited to the simpler modules that characterized the first 18 years of DEC's computers.

The two chapters in this part consist of a 1957 paper by Ken Olsen and a historical review by Dick Best. Both of these papers, but in particular the Olsen paper, give a glimpse of how early computer design was heavily weighted toward the electrical circuit level shown in Figure 1 of Chapter 1. As indicated above, the capability of modern technology to package complete switching circuit level and register transfer level systems into single chips has been a motivating force moving computer design toward the PMS level. There has also been increased activity "downward" however, as is also shown in Figure 1 of Chapter 1. To fit the modern, more complex systems into chips, increased attention to the lowest level (the device level) has also been required. Since this has been more the domain of the materials scientist than the computer scientist, it is not discussed in detail here.

While module design and computer design have evolved a great deal in the past 18 to 20 years, certain aspects of the Olsen paper reflect design methods which have counterparts today. In particular, convenient maintenance was plainly one of the important goals in the TX-2 circuit design effort. The use of a single, standard type of flip-flop and the use of a minimum number of different plug-in units were important elements in meeting that goal. These features simplified the design, simplified maintenance training, and reduced the variety of spare modules