Gregory's ServerThe ancestor of the 8086 processor is the Datapoint 2200, a desktop minicomputer used as an intelligent terminal. Made before the microprocessor, the Datapoint built a processor from a board of chips. The Intel 8008 cloned the Datapoint, first step to the x86 architecture. 🧵
 42 comments
The newfangled microprocessors were too slow, so Datapoint rejected them (bad move). Texas Instruments advertised their "CPU on a chip" but couldn't find a customer for the TMX 1795 and abandoned it (also bad move). Intel marketed the 8008, creating the microprocessor industry. Because the Datapoint 2200 used low-cost shift-register memory instead of RAM, it operated serially and needed to be little-endian. The 8008 copied this and that's why Intel processors are little-endian today. The Datapoint 2200 had a parity flag, very useful for a terminal. It had I/O instructions for its hardware. That's why x86 has a parity flag and uses I/O instructions. Intel improved the 8008 to create the 8080 processor, which was popular in embedded systems. The first generation of home computers (Altair, IMSAI) used the 8080. Because of backward compatibility, the 8080 still had the Datapoint instructions and features. The 8086 was a big improvement over the 8080, a 16-bit processor instead of 8. The 8086's registers names originally matched the Datapoint ones: A, B, C, D, E, H, L as shown in this 8086 patent diagram. But these were renamed AX, BX, CX, and DX just before release. The 8086 was designed to be backward compatible with the 8080 through a conversion program called CONV86, so it inherited the Datapoint features. The 8086 was extended to the modern x86 architecture used in most laptops and servers today. So that's how the modern x86 architecture developed from an obscure desktop computer called the Datapoint 2200. For lots of details and a close look at the instruction sets, see my blog post: https://www.righto.com/2023/08/datapoint-to-8086.html Credits: Altair photo by Colin Douglas, (CC BY-SA 2.0) https://commons.wikimedia.org/wiki/File:Altair_8800,_Smithsonian_Museum_(white_background).jpg  @kenshirriff I read your earlier (but in substance identical) account of the Datapoint/Intel history a while ago. It is mind-blowing how the design decisions of a rather obscure intelligent terminal in 1970 still shape a large part of computing today, more than 50 years later — and probably for decades to come. No one could have ever imagined that at the time, and I even have a hard time grasping it now. Thanks a lot for sharing this!   @kenshirriff @schotanus I have never worked on it but in the mid 80’s, Neddata, the IT part of Nedlloyd Shipping Company had one small but important system running on a Datapoint. The rest was working on IBM mainframe(s)and something new called DEC. We had one PC, for 125 it-emplyees. 😂 @kenshirriff I’d heard that the parity bit was brought forward from 4004 and it’s use in operating traffic lights. @alexr Unfortunately, there are two problems with that theory. The 4004 does not have parity and the 8008 is unrelated to the 4004.  @kenshirriff The fact that a design decision in the 1970s led to an architecture trait that has endured for over 50 years now is pretty amazing. @zorinlynx The IBM System/360 architecture from 1964 is similarly amazing, since IBM's mainframes are still compatible with it. @kenshirriff @zorinlynx  @kenshirriff BTW that looks like the video board - which used 14 shift memory ICs (7 bit ASCII, two chips per bit). This is a memory card from the d2200 recreation I’m currently working on. It’s not engineered for the power hungry 1405s which I’ll need to emulate with a daughter board. Decoder board is already done, processor board is at the layout stage. I’m adding as many blinkenlights as I can fit on 🙂   @kenshirriff That's the first time an explanation of why x86 is LE has made sense to me!  @kenshirriff AFAIK, TMX 1795 was never released to the public. But at least one chip _does_ exist, and it _does_ work. @kenshirriff This looks kind of similar to the bit-slice machine that I've been working with in my last university year (and later on). The most complex chip on that board was a 16x16 multiplier. Hence it was used as a DSP. @nwp The Intel 4004 and the Intel 8008 processors are unrelated apart from the part numbers. They have completely different architectures. Many of the same people worked on both chips, though. @kenshirriff I owned one of these in the early 1970's while living in a Philadelphia apartment. It was extraordinarily heavy, uppercase only (you could buy an extra board for lower case) and communicated over a 110 baud acoustic coupler.   @kenshirriff @kenshirriff @piikkiniska Your Datapoint looks older because it has the CTC logo instead of the Datapoint logo. Is it a Version I? @piikkiniska Very nice! The Datapoint 2200 Version I is pretty rare, at least I don't know of anyone else with one.   |
Here's the Datapoint 2200's processor board, crammed with simple TTL chips. Datapoint asked Texas Instruments and Intel if they could replace the board with a single chip. Texas Instruments created the TMX 1795, the first 8-bit microprocessor. Intel created the 8008 processor.