12

The Clueless Engineer YouTube channel posted a video yesterday exploring a Soviet K1810VM88 CPU he got from Ukraine.

Apparently he originally used the wording in the video title that the K1810VM88 was a clone of the 8088 but changed the wording to use implementation based on this comment:

They are not clones but rather functional equivalents. The Soviets did very little in the way of 1:1 cloning of electronic devices; they had very good engineers, and they preferred to perform reverse engineering and developing their own versions of western ICs rather than outright cloning them. I do know that these tend to run cooler than the originals, and the microcode may also be somewhat different, although the performance per clock is extremely close to the original...

So what's the definition of "clone" being used here that I'm unfamiliar with? Are they saying it's not exact and therefore not a clone? In that case would that mean we mostly use the word "clone" wrong when talking about PC clones?

Or is there some background left out and assumed understood, perhaps along the lines of reverse engineering not pertaining to clones because something else, perhaps "cleanroom"?

I'm clutching at straws here so what am I missing?

What are the technical definitions of clones and functional equivalents and what are the key differences?

10
  • 3
    Even clean-room reengineered IBM BIOSes were considered "clones". What the author of the youTube script did here is simply assigning a new meaning to an established term.
    – tofro
    Feb 16 at 12:16
  • 3
    cpushack.com/2021/01/26/the-story-of-the-soviet-z80-processor Here's some info on how Soviets had actually copied a chip. Feb 16 at 13:07
  • 2
    Later in the video, they compare the running temperature of the 8087 to the K1810 and conclude "it's a lot hotter" (the 8087) as if that weren't the case with the 8088.
    – tofro
    Feb 16 at 16:21
  • 2
    Looking at from where teh comment came, I'd say I'm a bit surprised about that the author changed the word clone to implementation at all. It doesn't seem to come from a very known source. To get a real answer about this, it might actually be better to throw the question over the language part of SE. For instance The Cambridge dictionary (dictionary.cambridge.org/dictionary/english/clone) has a special meaning of clone for computers: "a computer that operates in a very similar way to the one that it was copied from"
    – UncleBod
    Feb 17 at 5:56
  • 2
    A new comment under the pinned comment to the video. Someone giving a reason to why you could call it a clone. ("it is a clone. It is an exact cycle correct copy of 8088, including undocumented instructions, that cannot be reasonably implemented just following the available documentation...")
    – UncleBod
    Feb 19 at 20:52

3 Answers 3

25

"Clone" does not have a well-established meaning in electronics. What is almost certainly meant meant here is that the K1810MV88 is not a transistor-by-transistor copy of the 8088, but a functional replacement.

That's a collection of circuits that do the same job, in the sense of executing the same programs and producing the same results, but are not identical to the circuits in the 8088. There may well be significant resemblances between the two sets of circuits, but they would not be identical except by an extreme coincidence.

There are varying degrees of functional replacement, including:

  • The strictest case is a "drop-in replacement", where you can unplug one chip and replace it with the other (with the power off!) and everything works just the same. Since the Soviets used a different standard pin spacing on their chip packages to western practice (2.5mm rather than 0.1") they didn't usually produce drop-in replacements.

  • The next case is an electronic replacement, where all the signals coming out of the package behave the same, and the only difference is the pin spacing, and, potentially, pin arrangement.

  • If the signals are different, the chip is only software-compatible. This exists in varying degrees of strictness: there may be extra instructions, and the the execution timing of instructions may differ.

There are further intermediate forms of compatibility, depending on the quirks of the chips being replaced.

4
  • 14
    There were cases of actual layout copying, including non-functional structures which is how they were caught red-handed. Those for sure were ‘clones’ in my opinion. The rest is English language usage.
    – Jon Custer
    Feb 16 at 16:16
  • This is pretty much what I thought. But when the video creator agreed with the commenter I thought "well maybe I'm missing something". It seems pretty clear now that I wasn't. So it strikes me as a bit odd they both came to such a conclusion. I might post a link to this QA on the video. Feb 17 at 6:07
  • A simple adapter from 2.5mm to 0.1" would allow a true clone to "drop in", so that's splitting a very fine hair...
    – FreeMan
    Feb 18 at 23:09
  • @JonCuster: but that doesn't make much sense considering how we use clone (so the of X part. Nearly every hardware (and so also CPUs) has revisions and variants and we put in the same basket (original), and maybe different package. So if we would be so strict, most original CPUs are not even clone of the first 8086 produced. Feb 21 at 8:18
11

I'm clutching at straws here so what am I missing?

Maybe that use of the term 'clone' in most situations isn't following any hard definition but more some opinion. Sure, just looking at the term, it might mean a strict 1:1. Like copying something down to the finest detail. But at the same time we routinely talk about PC-Clones but mean machines made of different components in different configuration. All they have in common is the ability to run the same software ... more or less that is.

Do PC-Clones Exist?

While some early Taiwan clones may have been trace by trace copies of the IBM PC, already the very first Compaqs used different Layout, different PCB and of course different cases. Are they still clones, or just work-alike? What about higher speed, which heavily influences compatibility? What when different chip (sets) were used, integrating / emulating components - are those still clones?

Cloning of Chips

Same goes for chips. It starts that a work-alike might be made in a different process which always results in different fine structure, over different technology (like CMOS vs. NMOS) and different Layout all the way to mostly different designs only exposing the same electric and software behaviour.

Here already the very first step is an important one, as chip design always goes hand in hand with the process used. When Siemens, Harris or NEC licenced Intel chips from 8080 onward, they not only got some schematics and masks, but the whole production technology - like Intel's HMOS for the 8085 - only that enabled them to produce exactly the same chip.

In reverse this means that any perfect clone is always a licenced product, as recreating a process is extreme expensive compared to doing a work alike with your own process.

Words are Not Always Neutral

Maybe this is the point to also tackle that usage of 'clone' well often meant to deliver a certain negative impression. Like being an inferior product due missing original effort and/or quality. An intention that may outwight any technological reasoning what a clone is or not.

To Copy or Not

Compare Clone to Copy. The are often used interchangeably and equally bad defined.

Gyro Gearloose and the Ducks

In the end usage of 'clone' always comes down to the Duck-Rule:

If it walks like a duck and quacks like a duck then it must be a duck.

'like' being the keyword, as it includes a subjective ruling of features according to the observer. In case of he K1810WM88 nicely shown by plugging it into a PC with no noticeable difference in working. Quack Quack...

For 99% of all PC users (OCD variants excluded) that's all it needs to qualify as a 100% clone.


Generic Comments are Rarely a Good Source

They are not clones but rather functional equivalents. The Soviets did very little in the way of 1:1 cloning of electronic devices; they had very good engineers, and they preferred to perform reverse engineering and developing their own versions of western ICs rather than outright cloning them. I do know that these tend to run cooler than the originals, and the microcode may also be somewhat different, although the performance per clock is extremely close to the original...

This sounds to me quite like a Soviet-fan-boi comment intended to support a certain PoV without providing any evidence.

It was not only a clear order to do copies of western chips - as they did with other tech - and so they did. Or at least tried.

As mentioned they immediately hit the process barrier, as just getting schematics even on layout level doesn't help producing them if one doesn't have the whole process, including all physical and chemical steps. As a result any Soviet clone would be different on layout level. It would also most likely differ in components used - transistors are results of their process and so are their features which a circuit must take into account.

Does this qualify to make it a not a clone? Not in my book. Only if one tries hard to make it look like an independent development.

Also, there is no single 'Soviet Engineer' or even company/plant copying chips. Already within the USSR several competing institutes and manufacturers worked more or less independently, in addition to similar efforts in other COMECON countries.

The Z80 cloning by East Germany is well documented, showing the very same issues as mentioned. Still everyone, including its creators, regard the U880 as a clone.

19
  • 1
    @hippietrail maybe, impossible forme to judge. But if so, then using clone instead of licenced production seems would be rather misleading. Those chips were made with an explicit blessing of Intel, resulting in many years of easy profits for Intel. Not exactly what most would associate with 'cloning'.
    – Raffzahn
    Feb 16 at 11:07
  • 2
    @masswerk functional equivalent - that is a key. So a "glass terminal" is not a clone of a teletype because even though it is identical as far as signals, it is not functionally the same - a real teletype produces hard copy (which has advantages - printers have not disappeared even after 50 years of video terminals). I posit that a Compaq or similar can be argued as a clone if it has substantially all the same functionality, software compatibility (to any practical level), etc. even if it also has additional functions, higher speed, etc. In that vein, you could argue that a DECWriter Feb 16 at 15:01
  • 1
    In the end of the day, I personally consider "functional equivalent implementation" to be more neutral and "clone" more often a political term, the latter alluding to a low-effort, me-too product with no value added, attempting to occupy the same market space as the original that was cloned. (Some of these may not be true for the K1810VM88.)
    – masswerk
    Feb 16 at 15:34
  • 1
    @barbecue Nice, but then, as mentioned, next to no PC attributed as clone, wouldn't be one. Usage in computer thus seems to differ. After all, it's not unusual that one word may cover a different meaning depending on context.
    – Raffzahn
    Feb 16 at 21:33
  • 2
    @barbecue I think you're quite right. Terminology evolves (kinda like a driver was originally someone with a stick and a bunch of goats - not many associating that nowadays :)). But inbetween things that seems clear at first becomes extreme messy when looking close. Also, yes, this can often be traced to lousy usage in magazines, keen on compact headlines, often written by journalist not much more informed than their readers. On the long run we still need to improve that by making clear what is meant - and the very core of the question asked, isn't it?
    – Raffzahn
    Feb 16 at 22:55
5

I think what would distinguish a "clone" from something else would be whether its producer's objective was to make something that would be viewed as bidirectionally interchangeable, as compared with trying to make something that either sacrificed capability for price but would be "good enough" for most applications, or added capabilities beyond what had been present in the original in the hopes of making the product usable in ways the original wouldn't be.

Consider, for example, the NEC V20. It included hardware to process many instructions somewhat faster than the 8088, and a few instructions much faster, and also to emulate the 8080 used by CP/M systems. It was marketed as a cheap way for people with 8088-based PCs to improve the performance of their systems by replacing their existing 8088 processor, and also add compatibility with CP/M (though I think the speed improvement was a more popular reason for upgrading). By contrast, a company like AMD might have offered their chips to board makers as a cheaper alternative to Intel's 8088, but would have offered no reason for someone who has an Intel 8088 to replace it with AMD's product.

By contrast, many companies made IBM XT clones whose objective was to be indistinguishable from the XT on any meaningful basis other than price. Nearly all of them considerably reduced the time for a power-on memory test, and I don't think any licensed Microsoft's BASIC ROMs, but everything else about them was designed to be as thoroughly interchangeable with XT motherboards as possible.

12
  • 1
    An earlier edit had mentioned that about the V20; I hadn't meant to omit it. Thanks. It fits quite well my point that the V20 was trying to be better than the original, rather than being the same, but cheaper.
    – supercat
    Feb 16 at 17:31
  • Which reminded me of the DEC Rainbow, which had a Z-80 and an 8088, using the Z-80 as an I/O (floppy specifically) processor and also to run CP/M as an alternate to MS-DOS and other 16-bit operating systems. Just looked it up in Wikipedia and found "The 8088 could be upgraded to the faster NEC V20 chip. This gave about 10-15% speed improvement, but required changes to the system's ROMs to fix two timing loops." which is fascinating, because if the machine had been designed with the V20 in the first place it probably could have been built cheaper by getting rid of the Z-80 but still having Feb 16 at 17:46
  • CP/M compatibility, perhaps with a simpler floppy controller. Feb 16 at 17:47
  • 1
    AMD does not work as an example for cloning, as AMD licenced the 8088 it from Intel, including all tech needed. Back then they were still BFF, cross-licencing most of their chips. Only with the 386 Intel started to ghost AMD - and that's the point AMD reverse-engineered them
    – Raffzahn
    Feb 16 at 18:28
  • 1
    @Raffzahn: I'd refer to AMD as a "licensed clone"--the objective was to be bidirectionally interchangeable with Intel's parts. Clones may, but need not, make use of what would today be viewed as IP; the processor in the NES is an unlicensed clone of the 6502, with just enough circuitry disabled to avoid patent infringement.
    – supercat
    Feb 16 at 18:36

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .