10

The Harris company, well known for being an early second source making x86 chips under license, eventually produced a fully static CMOS 80C286 of their own design (later known as the Intersil 80C286 and apparently also sold by Intel). This processor was very popular in late PC AT clones on account of being able to run at a whopping 25 MHz.

Does the Harris 80C286 have any difference in behaviour from an Intel 80286 as far as the programmer is concerned? For example:

  • Do any instruction timings differ?
  • Are there any additional or different documented or undocumented instructions?
  • Does the Harris 80C286 microcode behave different from the Intel 80286 microcode?
  • Is it possible for a program to detect that it's running on a Harris 80C286?

Despite its popularity, very little details on the innards of this chip seem to be available on the internet.

1
  • The Harris device is available in a military version - that is probably a significant reason for it existing with Intel's encouragement. Nov 8 '20 at 2:55
7

Does the Harris 80C286 have any difference in behaviour from an Intel 80286 as far as the programmer is concerned?

The Harris/Intersil/Renesas 80C286 is, like its 80C86 predecessor fully compatible with its NMOS brother. In fact, it fulfils as well all electrical parameters (voltage, thresholds, lead timing), except of course current and thus power consumption.

It's a 100% compatible drop in replacement made with Intel licence and support (*1). They are not clones but second source.

The data sheets do not reveal any difference either.

Do any instruction timings differ?

No.

Timing differences happened only with compatible but independent developed/enhanced processors - like NEC's V20 extended the 8086 with 80186 elements.

Are there any additional or different documented or undocumented instructions?

No.

Does the Harris 80C286 microcode behave different from the Intel 80286 microcode?

Why would that matter?

Is it possible for a program to detect that it's running on a Harris 80C286?

No, unless there's a machine dependent way to measure power consumption.


*1 - At that point it might be important to remember that Intel did licence their designs (including all support chips and up to the 80386) to a wide range of manufacturers, Harris been one of them. In fact, Intel was eager to get others to licence and manufacture their chips, as their customers wanted to have a secure delivery. Not going into single-vendor lock-in was a major concern back then. Thus absolute identical behaviour was a basic requirement to allow drop in replacement.

11
  • 1
    If the microcode is different, it could behave different in edge cases. Have you tested that the behaviour is exactly identical? Datasheets like to claim a lot of things that are only half true.
    – fuz
    Nov 7 '20 at 18:54
  • 2
    The 286 is an incredibly complex processor with “miles of microcode” as the designers said. Especially with respect to protected mode, the behaviour of all the high-level features is very complex and often not evident from the contemporary documentation (though it did get better). I would be seriously surprised if the 80C286 would behave exactly identical in all situations, given that it's apparently an original design as opposed to a conversion of the 80286 design to CMOS. If you don't know for sure, it's fine. Maybe someone else has made experiments before.
    – fuz
    Nov 7 '20 at 19:12
  • 2
    What about undocumented opcodes? Nov 7 '20 at 19:13
  • 2
    @Raffzahn Do you have a source about the Harris 80C286 being such a 100% functionally identical design as opposed to Harris' own development? If yes, I can accept your answer.
    – fuz
    Nov 7 '20 at 19:55
  • 3
    I assume that Harris took the original NMOS design from Intel, and converted it to a static CMOS design while keeping all of the logical operation the same. That means the same microcode is used, and the same logic gates in every functional block… just with totem pole gates instead of depletion mode precharge drivers.
    – Chromatix
    Nov 8 '20 at 11:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.