NEC V20: The Original PC Processor Upgrade

In the early 1980s, there was the IBM PC, with its 4.77 MHz Intel 8088 processor. It was an unexpected hit for the company, and within a few years there were a host of competitors. Every self-respecting technology corporation wanted a piece of the action including processor manufacturers, and among those was NEC with their V20 chip and its V30 sibling. From the outside they were faster pin-compatible 8088 and 8086 clones, but internally they could also run both 8080 and 80186 code. [The Silicon Underground] has a look back at the V20, with some technical details, history, and its place as a PC upgrade.

For such a capable part it’s always been a surprise here that it didn’t take the world by storm, and the article sheds some light on this in the form of an Intel lawsuit that denied it a critical early market access. By the time it was available in quantity the PC world had moved on from the 8088, so we saw it in relatively few machines. It was a popular upgrade for those in the know back in the day though as it remains in 2025, and aside from its immediate speed boost there are a few tricks it lends to a classic PC clone. The version of DOS that underpinned Windows 95 won’t run on an 8086 or 8088 because it contains 8016 instructions, but a V20 can run it resulting in a much faster DOS experience. One to remember, if an early PC or clone cones your way.

Hungry for the good old days of DOS? You don’t need to find 80s hardware for that.

14 thoughts on “NEC V20: The Original PC Processor Upgrade

  1. Every self-respecting technology corporation wanted a piece of the action including processor manufacturers, and among those was NEC with their V20 chip and its V30 sibling.

    Not wrong, but I’d like to add that Japan had its own IBM PC counterpart, the NEC PC-9801.
    https://museum.ipsj.or.jp/en/computer/personal/0011.html

    So perhaps it was in NECs own interest to build processors for its own line of PCs, as well.
    Their first i8088 compatible was the NEC D8088D, I think.
    It was second-sourced from the intel design, I assume.

  2. I was on the PC design team at Amstrad UK back in the day, mostly doing software. We did CP/M+ work for the old Z80 crates as well as DOS.I had a PC1512 which I stuck a V30 in so that I could run the CP/M compilers in native mode. Worked quite well, but to be honest it was mostly for hacker cred as the Z80/8080 emulators available at the time weren’t half bad.

    1. Kia ora Vik! I was quite impressed with the PC1512/1640 when they came to NZ. Exceptionally capable machines for the price. I sometimes wish I’d never given mine away.

    1. The original article has more detail, specifically that this applies to the DOS-mode edit.com that shipped with Windows 95. Windows won’t run on 80186, but if you want a semi-decent text editor for DOS, you can get that.

      1. Most early 90s software was fine with 286 level processors, I think.
        So unless they checked for an 80286 explicitly, they also ran on 8018x and V20/V30.
        Protected-Mode and XMS was another story, of course.
        By late 90s, software sometimes had 80386 instructions mixed into.
        EMU386 can help to get them running on an 286 system.

        Btw, this may seem like planned obsolence at first, but there’s a logical conclusion.
        The 80286/80186 or V20/V30 do have an extended 8086 instructions set (8086-2 ISA) that is useful.

        It has useful instructions such as Output String to Port/Input String from Port,
        which accelerate i/o quite a bit (up to ca. 500 KB/s on an V20 in an XT!).
        https://web.itu.edu.tr/kesgin/mul06/intel/instr/outs.html
        https://web.itu.edu.tr/kesgin/mul06/intel/instr/ins.html

        Especially in games, VGA/VESA BIOSes and so on it made sense to use 80286 instructions forbest performance.
        Some may even had used 80386 instructions to beat the competition.
        By mid-90s it was all about multimedia and performance.

        The 80386 instructions in ordinary DOS programs were likely a mistake done by the compilers, though.
        By late 90s, the default compiler settings assumed an 386/486 likely.

        PS: As a workaround, there were so-called CPU accelerators.
        By late 80s, users of 8088-based PC/XTs could install an 8086/80286/80386 card that did upgrade the CPU.
        Such as Microsoft’s MACH 10/MACH 20, Orchid Tiny Turbo 286 or Intel Inboard 386.

        This was mainly intended to get demanding software such as DESQView, MS Windows or Windows/386 running smoothly.
        Or to get 386 specific features (emulated EMS via CEMM.SYS or Win/386).

        By early 90s, users of 80286-based PC/ATs could install CPU daughtercards such as MakeIt486.
        The AT 386 users could install an Cyrix 486DLC, by comparison.
        A bit later, the so-called Overdrive processors became popular, too.
        Such as the Pentium Overdrive (POD) for 486 systems.

        I hope that wasn’t to much talking of mine, I just find this fascinating to discuss. 🙂
        The PC history was full of hacky solutions.

  3. As far as I remember, the 80186 was not a 8086 evolution like the 80286, instead its a 8086 with extra peripherals intended for the embedded market. The company I worked for at the time (Spider systems ltd, cira 1990) used it for its range of ethernet bridges , routers and terminal muxes. I may have designed a wire speed local bridge hardware at the time with it (wire speed being 10Mbit/sec…)

    1. Yes, it was a microcontroller version of the 8086.
      However, it had things in common with 80286, too.
      For example, it could run EMU386 which uses INT 6 of 80186/80286, which is used to handle CPU exceptions.

      It also does calculate segment addresses in dedicated hardware rather than using the ALU like 8088/8086 do, if memory serves.
      So it’s a bit like an 80286 with Protected-Mode stripped, but integrated peripherals instead.

      Some DOS PCs used the 80186 as a main processor, also.
      The Tandy 2000, BBC Master 512, Siemens PC-D, RM Nimbus 186 etc.
      In principle, the 80186 was used as an early kind of SOC (system on a chip) before the term was coined.
      As an one-chip CPU+motherboard combo, so to say.

      The Tandy 2000 and RM Nibmus 186 are notable, I think, because they were used for MS Windows 1.x.
      The Tandy 2000 was available before the PC/AT and the fastest, most advanced x86 PC for a while.
      That’s why it was used during Windows development.
      Likewise, the RM Nimbus 186 was used in UK schools to run a Windows based environment.
      With actual MS Windows software written for it. In the mid-80s!
      Normally, Windows software wasn’t a thing until Windows 2.03 or Windows/386.

      Speaking under correction, though. I’m no expert here.

    2. Well there’s a blast from the past and I’d like to say thank you for my bonuses back then, I repaired many many Spiderport terminal dooberries, not that they were bad, jsut that they were popular with some of my clients and sufficiently expensive that they were worth repairing.

Leave a Reply to Eric CampbellCancel 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.