I have been a participant in the microcomputer world since 1976. In that time I have seen many micro architectures being rolled out and sometimes they fail to attract sufficient market share to remain viable. Here are some that I remember.
I still remember that gaudy ads in “Electronics” (once the premier magazine of the industry, from McGraw Hill) that announced the Clipper architecture from Fairchild. It was a 32-bit device in an era of 16- and 8-bit devices and found some application in workstations. It didn’t help that Fairchild was taken over by National Semiconductor in 1987 and the Clipper division was sold to Intergraph for their in-house use.
In the mid-70s there were only two CMOS microprocessors. If you wanted low-power operation you were restricted to the RCA 1802 COSMAC device (which I don’t regard as a flop) and the Intersil/Harris IM6100.The IM6100 was a 12-bit device when everything was 8 bits. As I recall it couldn’t easily fit with the 8-bit-wide devices (unlike the 16-bit devices that came later that used odd/even access to make up the 16 bits). It was designed to be compatible with the DEC PDP-8 minicomputer and even had the same instruction mnemonics. Perhaps Intersil had licensed this since Intel had already copyrighted the 8080 instruction set, forcing Zilog to create a whole new set of mnemonics for the Z80, a device that was a super-architecture of the 8080. Despite all the PDP-8 software, the IM6100 never went anywhere. When I searched for the history of the IM6100 I came across a reference from Jon Titus that there was another PDP-8 clone, the MP12 from Fabri-Tek. Apparently its legacy is the same as the IM6100.
Whilst we are dabbling in minicomputers, Data General introduced the microNova in 1977. I seem to recall it even got a mention in that wonderful book “The Soul of a New Machine” by Tracy Kidder. The microNova certainly didn’t provide a lifeline to Data General and the organization eventually failed.
The Transputer was launched with much fanfare in the early 1980s. It was aimed at the parallel computing market and even featured its own language, OCCAM, named for the 14th century English monk who is remembered for his Razor although the idea was not really his. Isn’t history wonderful? The name associated with a concept or invention is often not the actual first person to come up with it. But I digress. Despite its promise, the Transputer was also a flop.
In the 1990s, there was a move to RISC architecture and one proponent was Scenix who took a basic Microchip design, installed one peripheral (an 8-bit timer clocked by the system clock) and improved the process so that the chip (SX18, SX20, SX28) would run at up to 100MHz. The idea was that it was fast enough so that all the peripherals could be implemented in software. I must admit I bought into the philosophy until I actually completed a project. The devices were marketed by Parallax until recently discontinued, but they never really set the world on fire.
National Semiconductor tried its hand at many devices, the SC/MP (“SCAMP”), the IMP, COP and even a CMOS version of the Z80. They must have had some success, the 4-bit COP seemed to go into a lot of appliances, but when National cut its digital lines and spun out Fairchild, none of the processors survived.
When Intel brought out its 8086 16-bit family and Motorola offered the 68000, Zilog countered with the Z8000. The only thing I remember about it was that Steve Ciarcia developed a co-processor board for the IBM-PC that was described in Byte magazine. Anyone remember the Z8000 now?
Speaking of the 8086, what about the NEC V20/V30 which were supersets of the x86? That will also give us a segue into the mighty Intel’s mis-steps. Few remember that Intel invented the Digital Signal Processor, the i2920, but for some reason they never pursued that market. I am sure most of us do remember the iAPX432, the processor that ran ADA, but where is it now? The i960 was, according to Wikipedia, a success but Intel dropped marketing it “as a side effect of a settlement with DEC in which Intel received the rights to produce the StrongARM CPU.” The 960 was predeceased by the less successful i860. And for my last also-ran let us remember the MCS-96 family. It was originally developed for the automotive market and did find some success elsewhere for a while and I am not sure that it quite qualifies as a failure, but it certainly left no legacy.
It could be that some of these are still in production and that my impression is wrong. No doubt I will be corrected in the comments below. Please also add your recollections as well.