Ahoy, DECmate II the little PDP-8 that could

Now, that's a lot of word processing. But under the hood it's still at least PDP-8 adjacent, even considering its oddities and incompatibilities, and you can make it do many of the things a full-size Eight can. We'll take this basic unit, convert the floppy drives to solid state, tap the video output, and put it through its paces. After all, if we have a PDP-11 on our desk, we should really have a PDP-8 too.

Naturally the story starts with the PDP-8 itself, officially the seventh member of Digital's Programmed Data Processor family (the PDP-2 was never built). The PDP-8 is a 12-bit system that traces its lineage back to the 1961 LINC ("Laboratory INstrument Computer") developed initially at MIT's Lincoln Laboratory and manufactured by Digital and others. Also a 12-bit design, the 2048-word LINC was a capable and even beloved machine due to its flexibility and ease of interfacing, and it is generally considered the first effective "minicomputer" — not necessarily in size, mind you, but rather as a simpler architecture and reduced instruction set as opposed to larger systems. It was nevertheless fully comparable with them, and DEC systems architect Gordon Bell and designer Alan Kotok (later co-founder of the W3C) developed a simplified specification based upon it aimed at smaller market applications where the 18-bit PDP-1 and PDP-4 would have been unattractively priced. This became the 1963 PDP-5, its logic principally designed by engineer Edson de Castro (later the founder of Data General). It started at $27,000 [in 2026 dollars about $282,000], a surprisingly low cost for the era, and about a thousand were sold.

Castro expanded the PDP-5 further, mindful to retain the easy interfacing that made it popular, but at the same time making it faster and cheaper through parsimonious design and various architectural and microcode improvements. He was also able to make it smaller: built out of diode-transistor logic on closely packed flip-chip modules, it was now merely the size of a small refrigerator. DEC introduced it as the PDP-8 minicomputer in March 1965, as shown here in its original form at the Computer History Museum, for an even lower price at "just" $18,500 [$190,000]. It provided 4096 words of magnetic core memory and ran at a cycle time of 1.5us, yielding an effective clock speed of 667kHz. These initial models were later nicknamed the "Straight-8" and became the best selling computer model to that time, with nearly 1,500 rolling out to customers. The PDP-8 went through a rapid stepwise process of evolution, making it even cheaper and more capable, in addition to the side-branch LINC-8 and PDP-12 which specifically supported LINC instructions for those customers. These improvements arguably culminated in the 1970 PDP-8/E, a versatile system supporting up to 32kW of RAM that a year and a half and a price cut later became the first computer to sell for under $5,000 [$4,995, or in 2026 dollars around $52,000]. Although ostensibly a consolidated re-design of the TTL-based 1968 PDP-8/I, its stronger I/O capabilities using the new OMNIBUS and a wide variety of peripheral and system options nevertheless enabled it to scale from small installations up to very large ones, in a variety of settings (this particular unit assisted monitoring during neurosurgery at Massachusetts General Hospital). DEC offered several operating system choices as well, ranging from the simple PS/8 executive and its descendant operating system OS/8 to the multiuser TSS/8 supporting up to 16 users.By the mid-1970s, however, the PDP-8 was finally showing its age. The 1974 $1835 PDP-8/A had reduced it to a single board and expanded total memory even more, supporting both classic core memory and new semiconductor RAM, but although it found use as a lower-end option for office tasks (such as the 1975 DEC Datasystem 310) and basic CNC automation, compared to newer systems its architectural idiosyncrasies were increasingly seen as a liability. Exceeding the fundamental 4kW addressing limit required grafted-on registers, only addresses in the current 128-word page or the zeroth one could be accessed without indirection, the small basic instruction set made excessive use of "magic" locations and sometimes needed cumbersome code sequences for simple tasks (e.g., logical-OR), and its subroutine call storing the return address in the subroutine's first word inhibited efficient recursion — or running from ROMs. While Digital intended to keep selling Eights as long as there were buyers, at the time the corporation saw little profit in evolving the platform further.

In the meantime, few (if any) patents and copyrights had persisted on the PDP-8 instruction set or its major design features, to some degree because the LINC from which it was descended was government-funded and in the public domain, and combined with its simplicity and enduring popularity it became an attractive target for clones. A couple even appeared during the architecture's commercial lifetime — somewhat to Digital's annoyance — such as the Digital Computer Controls DCC-112, introduced in 1970 as a faster clone of the PDP-8/I. Even as DEC tried to wind down the line, its residual presence nevertheless remained sizeable, and other clones like the 1974 Fabri-Tek MP12 emerged to service the market. Despite Fabri-Tek billing the core unit as a "microprocessor," the main CPU module was still all discrete TTL.But the first commercial microprocessor implementation of the PDP-8 didn't come from DEC either — in fact, for a period of time DEC under co-founder Ken Olsen actively repulsed internal attempts to do so. As we talked about before with the DEC Professional, the emergence of the microcomputer became an existential crisis to minicomputer makers, including DEC but also HP, Data General, TI, IBM and others, and many attempted to compete by shrinking down their current offerings (such as TI turning the 990 into the TMS9900, DG with the ill-starred microNOVA, and the HP 2100 as the Binary Processor Chip). An DEC skunkworks project did the same to develop what was intended as the PDP-8/B in 1973, but management was sceptical of the investment required and it was cancelled. Shortly afterwards, in 1974, developers squeezed a reduced PDP-8/A logic board into a VT50 terminal and demonstrated it as one of two potential personal computer products to Olsen. To their disappointment (including a young David Ahl), he vetoed them also on the advice of management concerned it would cut into existing product lines, making the infamous observation that no one would want a computer in their home. (Olsen would repeatedly attempt to clarify this statement, even doubling down on it in 1977, citing the size and inconvenience of then-contemporary computer systems.)

The installed base of PDP-8s was still significant, however, and upstart semiconductor company Intersil saw an opportunity of their own. Intersil was founded in Cupertino by Swiss physicist Jean Hoerni in 1967, one of transistor co-designer William Shockley's "traitorous eight" who left Shockley Semiconductor to form Fairchild in 1957. At Fairchild Hoerni patented the planar process of semiconductor production in 1959, still an elemental concept even today in modern photolithography, then going on to found Amelco, later Teledyne, in 1961 and consult for the new Union Carbide Electronics in 1964 before establishing Intersil. Intersil became an early pioneer in low-power CMOS through their work in watch timekeeper chips, where power saving was critical, and were able to fabricate full CMOS designs years before many competitors.

Intersil's 1975 development of a pure CMOS PDP-8 CPU was completed independently of DEC, intending not only to create a microprocessor of their own but also one with a presumably guaranteed market: the well-known instruction set would enable it to run much existing software, but could run up to 4MHz (from an 8MHz crystal), and its fully static CMOS design and unc