The DEC PDP-8 family of computers is most significant not so much because of a novel architecture but rather because of the fundamental shift in the mode of computing that it represented. It defined the class of "minicomputers"-machines which were cheap enough for an individual research lab to own, and small enough to be embedded in other devices. In fact, PDP stood for "Programmed Data Processor;" because it was such a break from "traditional" computers that Ken Olsen decided against calling it a "computer." As Olsen put it, existing computer users "could not believe that in 1960 computers could do the job could be built for less that $1 million" (Aspray 224). PDP-8's were marketed to the scientific research community with the expectation that each would be extended and programmed by the customer to suit their needs, a stark contrast to the commercial solutions which IBM leased to its customers.
The PDP-8's small size and cost were made possible by the extensive use of transistor-based and, later, integrated circuit-based logic circuitry, as well as a reduced word size (12-bit) and the absence of the multitude of peripherals present in larger "mainframe" computers (Ceruzzi 129-32). Because of this shorter word length, the PDP-8 could not compete with IBM's mainframes in doing arithmetic on ten-digit numbers or floating-point numbers (Ceruzzi 132), but from the start it was "designed for task environments with minimum arithmetic computing and small [memory] requirements" (Computer Structures 120). Thus the PDP-8 was thus not aimed at the commercial data-processing market but the science and engineering market. These customers were technically advanced-they could write their own software, they didn't need advanced peripherals, and they could develop their own I/O devices to interface to their existing equipment.
The compactness of the architecture was not without its quirks when it came to programming. One of the most difficult problems was the need to address large amount of memory from within a 12-bit instruction, of which 3 bits were already used for the opcode. The solution was to use the remaining nine bits to address a word in the current 128-word memory "page" or page zero. An entire 12-bit address could be created by using "indirect" addressing mode, in which the memory address referenced by the opcode contains the 12-bit memory address of the location that the program actually wants. This situation was further complicated when the computer needed to expand beyond 212=4096 words of memory. The extended memory unit added the concept of the current "field" of memory-one of eight 4096-word ranges where instructions and data could be stored. (see ch 8 of Computer Engineering and Doug Jones' online reference) One interesting effect of this use of local addressing is that data and code become extremely intermixed, as the JMS instruction (jump to a subroutine) actually places the return address in the memory address it references and continues to the next instruction. Thus the beginning of each subroutine is a zero; this location will be filled in with the return address as the program executes. It is also worth noting that the instruction set leaves a wide range of instructions defined simply as I/O instructions, some of which applied to existing DEC hardware peripherals, such as the teletype, but many others which could be used to address custom-built hardware on the PDP-8's I/O bus.
The accessibility, in size and price, of the PDP-8 family gave many their first hands-on use of a computer. No longer would MIT have only one IBM mainframe, on which students would submit cards and receive the results of the computation-or the errors generated by them-as another stack of cards. This change had in fact already begun at MIT with the use of an earlier DEC model, the PDP-1. The difference between the use of the MIT PDP-1 and MIT's IBM mainframe is striking; in fact, one of the PDP-1's most famous applications was for the Tech Model Railroad Club's layout. With such a seemingly frivolous use of what would then be considered valuable computing power, we see that "clearly the economics of mainframe computer usage, as practiced not only at commercial installations but also at MIT's own mainframe facility, did not apply to the PDP-1" (Ceruzzi 129). As tens of thousands of PDP-8 family computers flooded the research community, many students, technicians, and seasoned engineers had their first hands-on experience with a computer, in the full range from programming to developing custom hardware for the computer. A strong hobbyist culture arose from these users, leading to the creation of games for the machine. These hobbyists often referred to the minicomputer as their "personal" computer, (Aspray 223) further indicating the evolving relationship between the user and the machine.
It is worth noting that the concept of personal interaction with the machine was also present (albeit not in the physical interaction) in the time-sharing systems developed during the 1960's. This was in fact part of the idea behind the vision of the "computer utility," in which centralized time-sharing machines would be used by numerous users with remote terminals. This vision was actually coming to fruition in the late 1960's, as around twenty firms, including such giants as IBM, BBN, and General Electric competed for the $15-20 million market of time-shared computing. This vision was in part propelled by "Grosch's Law," which stated that the computing power of a computer varied as the square of its price. This clearly made it economical for a large number of users to utilize the power of a central machine. However, the advent of the minicomputer, in the form of the PDP-8, shattered Grosch's Law by so drastically reducing the price of computers that it actually made sense, if time-sharing was required, to buy a minicomputer and set it up to accommodate perhaps a dozen users instead of subscribing to a commercial time-sharing system (Aspray 215-22).
The small size of the PDP-8 actually allowed it to be built into other devices, thereby originating the term "OEM" for "Original Equipment Manufacturer." Such uses for the PDP-8 included the DEC 388 display computer (Computer Engineering 201), the LS-8 theatre lighting system from Electronics Diversified (which played a key role in the success of the Broadway hit A Chorus Line), and even a potato-picking machine installed on a farm tractor (Ceruzzi 135-6). Here, too, the small size and price of the machines in the PDP-8 family made possible a shift in the use of computing power. Certainly no one would consider using an IBM mainframe to control theatre lights or to pick potatoes. However, the size and affordability of the PDP-8 made computer power cheap enough for such uses to be possible, practical, and even marketable.
These notions of personal interaction with a computer and the use of computer power in other devices has certainly carried through to the present day. The desire on the part of computer hobbyists in the 1970's to have their own computer helped fuel the development of the personal computer, and computing power is now to ubiquitous that we don't even notice that an elevator remembers who pressed a button or that a digital alarm clock can remember what time to wake us up. The advent of the minicomputer, and its popularization in the form of the PDP-8, brought about the fundamental shift from the model of a few centralized computing machines to the notion that computers would be directly accessible to its users and could be embedded in other devices for tasks which are not explicitly computational. These trends have certainly had a strong-perhaps defining-impact on the current uses of computers..
Aspray, William; and Martin Campbell-Kelly. Computer: A History of the Information Machine. BasicBooks, 1996.
Bell, C. Gordon; and Allen Newell. Computer Structures: Readings and Examples. McGraw-Hill Book Company, 1971. (also available online)
Bell, C. Gordon; J Craig Mudge; and John E. McNamara. Computer Engineering. Digital Press, 1978.
Ceruzzi, Paul E. A History of Modern Computing. MIT Press, 1998.
Tim Gorton May, 2000