Making Intel Mad, Retrocomputing Edition

Intel has had a deathgrip on the PC world since the standardization around the software and hardware available on IBM boxes in the 90s. And if you think you’re free of them because you have an AMD chip, that’s just Intel’s instruction set with a different badge on the silicon. At least AMD licenses it, though — in the 80s there was another game in town that didn’t exactly ask for permission before implementing, and improving upon, the Intel chips available at the time.

The NEC V20 CPU was a chip that was a drop-in replacement for the Intel 8088 and made some performance improvements to it as well. Even though the 186 and 286 were available at the time of its release, this was an era before planned obsolescence as a business model was king so there were plenty of 8088 systems still working and relevant that could take advantage of this upgrade. In fact, the V20 was able to implement some of the improved instructions from these more modern chips. And this wasn’t an expensive upgrade either, with kits starting around $16 at the time which is about $50 today, adjusting for inflation.

This deep dive into the V20 isn’t limited to a history lesson and technological discussion, though. There’s also a project based on Arduino which makes use of the 8088 with some upgrades to support the NEC V20 and a test suite for a V20 emulator as well.

If you had an original IBM with one of these chips, though, things weren’t all smooth sailing for this straightforward upgrade at the time. A years-long legal battle ensued over the contents of the V20 microcode and whether or not it constituted copyright infringement. Intel was able to drag the process out long enough that by the time the lawsuit settled, the chips were relatively obsolete, leaving the NEC V20 to sit firmly in retrocomputing (and legal) history.

20 thoughts on “Making Intel Mad, Retrocomputing Edition

  1. “that’s just Intel’s instruction set with a different badge on the silicon. At least AMD licenses it, though”

    In the same way Intel licenses AMD64 instructions in their chips. So it’s about a 50:50 share under either badge.

    1. March 2004 unveiled the “official” name EM64T (Extended Memory 64 Technology). In late 2006 Intel began using this name instead of using the name Intel 64 for its implementation, paralleling AMD’s use of the name AMD64.

  2. > At least AMD licenses it, though

    … which it shouldn’t, since that tends to reinforce the false assumption than an ISA is copyrightable (or patentable, or anything else that requires a license). This is, of course, an assumption that probably suits AMD almost as much as Intel.

    > A years-long legal battle ensued over the contents of the V20 microcode and whether or not it constituted copyright infringement.

    … which it obviously did not from day one, but Intel’s legal harassment still worked.

    > Intel was able to drag the process out long enough that by the time the lawsuit settled,

    It wasn’t settled. Intel lost completely. It was just too late.

    1. AFAIK, the license includes the software tests to guarantee that an implementation honors the ISA. Creating those tests is very hard, and Intel/AMD have developed during the years a lot of tests for specific cases, tests that are very valuable to ensure that your design works as expected in all the cases. If you read “The soul of a new machine”, there they talk about these kind of tests.

  3. What’s also important maybe:
    The NEC V20/V30 found use in NECs own PC line, which was popular on the Japanese island and Intel had no saying here.

    In essence, there had been PC-6001, PC-8001, PC-8800 (PC-88) series and the 16-Bit PC-9801 (PC-98) series.

    And the early models ran on Z80 CPUs, prior to PC-98.
    The PC-9801 was the Japanese counter part to the IBM PC 5150.

    It ran on 8086 processors originally, but some models had an NEC V30 or 80286.

    Now, adding a Z80 or V30 in addition to the main CPU gave some PC-98 machines limited backwards compatibility with PC-88 software.

    I know, my wording is very poorly done here. Sorry about this.

    What I mean to say is that the V20/V30 was more than an upgrade chip for an IBM PC or compatible.
    It also features 8086-2 instruction set (80186/80286 level) and 8080 emulation mode.

    The 8080 emulation mode could have been useful to NECs own computer line, it’s not necessarily designed as a CP/M feature.

    Maybe old software (PC-88 etc) didn’t use Z80 specific instructions,
    if being compiled with the contemporary compilers available to the Japanese platforms.

    Anyway, these are just some thoughts.
    Japanese companies traditionally focused on their home market first before expanding into foreign markets.

    So maybe we, the Western market, weren’t as important as we thought we would be.
    Maybe we should try seeing things from other’s perspectives more often.

    1. NEC had another model CPU, the μPD9002, which was x86 compatible but included the Z80 instruction set instead of the 8080 instruction set. They used these for backwards compatibility on the PC-88 line.

      1. I know, this was a very fascinating processor. Glad you mentioned it!
        Unfortunately, it was used in merely one model or so (the PC-88 VA and two more variations of it).
        If the processor only had been available for sale overseas..

        In Europe, we had quite some 8086 PCs that could have used it well.
        The Amstrad/Schneider PC1512 and PC1640, Olivetti M24 or IBM’s PS/2 Model 30 (8086 ver) and so on.

        Btw, some PC-98 models had an V30 and a higher-end processor, such as 80286.
        I’m not entirely sure if the additional V30 wasn’t also used for compatibility with old PC-98 models.

        This site has a chart:

  4. The planned obsolescence comment and the newer features on old form hade me reminiscing about the Cyrix 5×86-133 chip and adapter for upgrading a 486 system to arguably low-end Pentium performance though at 1/2 the bus speed. Was way cheaper though.

    1. The Cyrix 5×86 and 6×86 were arguably 486’s with Pentium instructions added to them them. They performed poorly compared to a real Pentium especially the FPU

      1. The other contender, the Am5x86-PR75, which everyone i knew ran overclocked, was a good contender tough. It ran, as advertized, as fast as a Pentium 90 for a fraction of the price, i even enjoyed Quake 1 on it back then. 320×200 but it ran and that resolution was considered playable back in the 90s.

  5. The V80 also implemented the 8080 instruction set, allowing an IBM PC to run CP/M 80. Unfortuantely, the Z80 was king of the 8 bit intel derived world by then, so some popular software wouldn’t run.

  6. I have three “western” vintage machines one packard bell one Italian and one pc jr with upgrade board. These all are using v20 chips. Only the original IBM has the 8088 in my collection. From my viewpoint V20 has been used quite a ot.

    1. Well, it was and still is a nice and cheap upgrade for an XT class machine. My Tandy 1000 has one inside and it is a nice little boost. Didn’t even break the bank, i think i paid somewhere around 10€ for it.

      1. The V20/V30 also enhanced compatibility with 90s era software.
        Most DOS software compiled with an 80286 compiler switch would run on V20/V30 PCs.
        Later versions of PC GEOS could be made run with it, if memory serves.

        Same goes for Turbo Pascal for Windows applications.
        Windows 3.0 runs on 8088/8086 in Real-Mode, but many programs require one of the two Protected-Mode kernals.

        Those who’re Real-Mode friendly may use 8086 or 80286 instructions, depending on the compiler flag used.
        With a NEC processor, both should run.
        (Lander 3 and GNU Chess run on plain 8088/8086, for example.)

        Also interesting is the fact that Windows 3 VGA driver depends on 80286 instructions.
        It won’t run on 8088/8086 PCs unless being patched (there’s an 8088/8086 patch at vcfed forums).
        With a V20/30, the non-patched version can be used. With slightly better performance, I assume.

  7. All comments fun to read.

    But ARM M and A and RISC-V low-cost and low-power platforms
    hosting a Linux [or 256 bit Intel MCS BASIC-52, of course? :)] which runs the gcc c compiler the current future?

    “AMD CEO Lisa Su spoke at the ITF World 2024 conference, where she outlined the company’s plan to achieve 100x power efficiency improvement by 2027.

    Power, not latency, to guide computing at the edge.”

  8. isa is arcane 1943 tech that Motorola invented in 1943. No, Intel isnt borrowing from amd. They, have the true 64 core Itanium witth each core True 256bit. So they dont need to borrow from pos isa true 8bit amd x86 as are all x86. They have to live it. They wont. its called Hindering You the public..They, are illuminati.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.