I think the 20mhz part should work in any board that can support a 20MHz bus.

That 20 MHz bus is more common in laptops methinks. I have an NEC Versa V 40EC, and it runs a 486 SL DX2 40MHz....clock double means 20MHz bus speed. That thing is surprisingly nimble for such a slow system bus for a 486. Even runs some heavy stuff at seemingly full speed given a lot of RAM (Diablo and NESticle to mention 2).

Scali wrote:

mpe wrote:

In "my area" I hardly remember seeing a 386 chip that was made by Intel.

That would be very odd, given that Intel has been supplying 386 CPUs since 1985, and AMD only from 1991 onwards.

Well, I got my first PC in 1991. Perhaps that's why 😀. I was a 8bit user before that. However, I suspect I’d see more 286's than 386's before 1991.

Also PC volumes in 80's that were nothing compared to 90s. From my memory I remember 3rd party 386 variants was an instant hit. You could get more megahertz for less money. In at least partially budget sensitive situation, this was no brainer. If you were buying from OEMs or went to 486 nobody did!) it could be different.

Scali wrote:

Those aren't clones. Those are second-source Intel CPUs.

Which Intel chip is the Intersil/Harris 16/20/25 Mhz 80C286 a second sourced version of? These were clone chips based on scaled up from Intel's original license. They directly competed with Intel’s 386SX. The formed "second-source" partnership quickly become fierce competition once Intel rejected to grant 386 rights to former licensees. Intel's fastest 286 was 12.5 MHz. Perhaps the last one that could be second sourced (or perhaps 8 MHz?).

It is very similar to Am386 which also wasn't a clean room design, but a scaled up version of the 386. Intel claimed licence was for 8086/80286 while AMD thought it's for x86. And that was the reason of litigations (or one of many).

Scali wrote:

. […]
Show full quote

.

ShovelKnight wrote:

A fraction of those who bought the "lesser" chips would also buy a 487 in the future, which drove additional profit for Intel.

A very VERY small fraction, which probably wouldn't have made any kind of dent in Intel's sales figures whatsoever.
Virtually no software required an FPU in the 486-days, so there was also no reason to upgrade. Apart from the fact that the main reason why people would get a 486SX instead of a DX was to save money. Getting a 487 doesn't re ally fit with that market.

All the 486SX systems I have aquired over the years were indeed upgraded to DX2 class 486s. It made sense to upgrade if you could without having to swap out the whole base. This includes DECpc, ACER, IBM and Compaq systems I have. 94-95 486DX2/66 systems were still being sold by the bucket load. It didn't take long to realise you didn't need a special "487" for that upgrade made things more compelling. AM386DX-40s and 486DLCs filled quite a nice little niche. They were quite a big seller in our neck of the woods in early 90s. Indeed out selling early OEM 486SX systems. Folk bought systems with 386DX-40s/486DLCs which included 486 upgrade sockets on the mobo, of which I have an example and the original purchase paper work. Considered doing the same thing myself but decided not to as my 286/16 was doing all I needed to do at the time and couldn't justify the $800 outlay. The system was first bought with a 486DLC33 cpu originally then when Windows 95 came out upgraded the cpu to a 486DX2/66, upgraded his ram and added an extra hdd. Of course I received it in the later condition with the original DLC33 cpu which was nice. You didn't replace your system every two years back then.

As for my 286. A freind upgraded his 486 to an early Pentium, always one for keeping up with the latest and greatest, handed down his mobo for a princely sum. I repurposed my SIPPs in that by removing the pins and had a nice Dos/Win3.1 rig. All fitted in a flip top mini AT case.

mpe wrote:

Well, I got my first PC in 1991. Perhaps that's why :). I was a 8bit user before that. However, I suspect I’d see more 286's tha […]
Show full quote

Scali wrote:

mpe wrote:

In "my area" I hardly remember seeing a 386 chip that was made by Intel.

That would be very odd, given that Intel has been supplying 386 CPUs since 1985, and AMD only from 1991 onwards.

Well, I got my first PC in 1991. Perhaps that's why 😀. I was a 8bit user before that. However, I suspect I’d see more 286's than 386's before 1991.

Also PC volumes in 80's that were nothing compared to 90s. From my memory I remember 3rd party 386 variants was an instant hit. You could get more megahertz for less money. In at least partially budget sensitive situation, this was no brainer. If you were buying from OEMs or went to 486 nobody did!) it could be different.

Scali wrote:

Those aren't clones. Those are second-source Intel CPUs.

Which Intel chip is the Intersil/Harris 16/20/25 Mhz 80C286 a second sourced version of? These were clone chips based on scaled up from Intel's original license. They directly competed with Intel’s 386SX. The formed "second-source" partnership quickly become fierce competition once Intel rejected to grant 386 rights to former licensees. Intel's fastest 286 was 12.5 MHz. Perhaps the last one that could be second sourced (or perhaps 8 MHz?).

It is very similar to Am386 which also wasn't a clean room design, but a scaled up version of the 386. Intel claimed licence was for 8086/80286 while AMD thought it's for x86. And that was the reason of litigations (or one of many).

386DXs weren't selling as great as Intel wanted. Weren't really that many folk using 32 bit software to start with and the things were damn costly. Hence the introduction of the 386SX line. Cheaper to produce but with that "386" brand. No one cared it could only address 16 megs of ram, which was a HUGE amount at the time any way, at least it was 1 number ahead of the 286 and was 32-bit internally.

A fellow dropped of a 1989 386DX33 system, with FPU, he use to use back then for surveying under my car port when I wasn't at home one day As a proper work hack it was pretty shonkily assembled and had seen good use for a decade or so before he put it in storage. AT form factor that took 32megs but 8megs fitted. Unfortunately the Apple branded 100meg scsi drive went tits up before I had time to retrieve the software. Still going strong though with it's new 4gig BigFoot drive. Couldn't imagine the going rate it was sold for back then.

I dug thru my old motherboard docs and found one that listed a 486SX-16 cpu. Is an early rev of a Unisys 300 486 series board, the later rev of same supported 25 and 33mhz clocks and was about 6months later from the docs rev date. Uploaded it to the driver library here:
http://www.vogonsdrivers.com/getfile.php?file … &menustate=27,0

I bought my 286 in 1988 and I upgraded motherboard, cpu and ram in 1992 to 386dx 33Mhz. I kept all other parts of my 286 I skipped 486 completely and I built a new pentium 90Mhz pc in 1995 with windows 95.
They were selling 286s, 386s and 486s pretty much at same time, so people could have skipped any of them based on what they could afford back then.

Baoran wrote:

I bought my 286 in 1988 and I upgraded motherboard, cpu and ram in 1992 to 386dx 33Mhz. I kept all other parts of my 286 I skipped 486 completely and I built a new pentium 90Mhz pc in 1995 with windows 95.
They were selling 286s, 386s and 486s pretty much at same time, so people could have skipped any of them based on what they could afford back then.

Hell yeah XT class systems and C64s were still being sold new back then 🤣. Ahh 1990 The year is 1990 and you still have a PC/XT and can't upgrade. Which games are you playing?

mpe wrote:

Which Intel chip is the Intersil/Harris 16/20/25 Mhz 80C286 a second sourced version of?

The Intel 286 obviously. It may be clocked higher, but it's the same design, and licensed from Intel.
Higher clockspeeds are just the result of advances in manufacturing.

mpe wrote:

It is very similar to Am386 which also wasn't a clean room design, but a scaled up version of the 386. Intel claimed licence was for 8086/80286 while AMD thought it's for x86. And that was the reason of litigations (or one of many).

It's not similar at all. Intersil/Harris actually licensed the 286 and obtained the chip designs from Intel, so they are legal.
Intel never licensed the 386, and as such it did not give any chip designs to any third party at all. AMD had to reverse-engineer the Intel 386 and create their own (illegal) copy of the chip.

As said, Intel won that claim in court, the license indeed WAS for 8086 through 286 only, it did not cover the 386 or newer chips. That is a fact.

Then Intel tried to stop clean room implementations and that failed in the courts hence the fantastic 386DX-40 which sold by the bucket loads. http://redhill.net.au/c/c-4.html

Scali wrote:

It's not similar at all. Intersil/Harris actually licensed the 286 and obtained the chip designs from Intel, so they are legal.
Intel never licensed the 386, and as such it did not give any chip designs to any third party at all. AMD had to reverse-engineer the Intel 386 and create their own (illegal) copy of the chip.

It is funny when you say "Intel never licensed" and "create their own (illegal) copy". You seem to still hold the Intel's position from 1991 😀 AMD/other vendors opinion was obviously very different.

You forget that Intel's claim has been rejected by court in 1992 when the court ruled against Intel and ordered them to pay $10M plus a royalty-free license to any Intel patents used in AMD's 386. This was later forged into a settlement with even wider cross-licensing. In this perspective it was properly licensed from the beginning and wasn't an illegal copy, although Intel might have thought otherwise at some point in the history.

Taking this into the account the situation with the 386DX-40 is almost identical to those high-clocked 286 chips. Both started as licensed clones (although unlike 286 Intel was obviously not happy and cut access to many engineering resources once found they are working on it). In the end we've got a chip that was still based on Intel's 386 technology, used their microcode, but scaled up to significantly higher performance levels so that it was competing with Intel's more advanced design.

But we went quite far in the discussion. My original point was that I think that I think that main reason behind 486SX was the commercial pressure (low cost competition and slow 486 sales). More so than desire to recycle defective 486DX units...

Caluser2000 wrote:

Then Intel tried to stop clean room implementations and that failed in the courts hence the fantastic 386DX-40 which sold by the bucket loads. http://redhill.net.au/c/c-4.html

As I said, they didn't fail.
However, court ruled that Intel had to offer x86 licenses.
Since the 386DX-40 was under a cross-license agreement, it was no longer illegal in 1991 (after AMD replaced the illegally copied microcode with a clean-room reimplementation). Before that, it was, and it was never sold (AMD had been trying to sell 386 clones since 1987).

By the way, NEC was also sued for microcode in the V20/V30, and their code was not considered illegal, because they had made significant changes to the microcode for their specific CPU.
So it was not copy-paste as the Am386 was.
Also, the court ruled that because the V20/V30 were derivations of the 8088/8086, the patent cross-license applied here (NEC was also second-source for regular 8088/8086 and various other Intel chips).
So the V20/V30 was eventually ruled as legal (which would certainly make the Intersil/Harris 286 CPUs legal, as they differed only in clockspeed, where the V20/V30 actually have an improved architecture, extra instructions, and modified microcode).

mpe wrote:

You forget that Intel's claim has been rejected by court in 1992 when the court ruled against Intel and ordered them to pay $10M plus a royalty-free license to any Intel patents used in AMD's 386.

No, that is simply not true.
Intel's claim was never rejected.
The court just used the argument that x86 had a monopoly position, and therefore needed to be opened for licensing.
That is something completely different from Intel's claims that AMD's 386 was unlicensed and therefore illegal.

After all, if the 386 was in fact legal, then there would be no need for any kind of license, would there?
So Intel was right about the status of the license (the patent cross-license did not extend to the 386, only 8086 through 286) and the illegality of the 386 (infringing on Intel patents, copyright and whatnot). AMD just used anti-monopoly tactics to get what they wanted. They just used the fact that Intel was so big and x86 was so successful against Intel. That's different. Get your facts straight.

If you want more details, AMD believed that Intel had to open negotiations for extending the second-source licensing to the 386 (by which AMD implicitly acknowledges that they do not have these rights). Intel countered that the terms of the agreement state that it is a *cross*-license, so Intel should also receive IP from AMD in return. Therefore, the agreement says that extension is only an option when Intel receives IP of "sufficient value" in return. Since AMD never did anything other than just selling clones of Intel parts, Intel never got anything in return from the cross-license. Therefore Intel never thought there was any reason to negotiate over the 386, and Intel eventually terminated the cross-license (which was also in the terms of the cross-license: either party could terminate after 5 years, with a 1 year notice).

So AMD basically argued: "But that's not fair! We have products, Intel just doesn't want to take them!"
But really, what product would AMD have that was actually of "sufficient value" to Intel, compared to the 386, which was a huge leap forward in the x86 world at the time? The "unfair" part was only honoured because Intel had a monopoly position.
Intel had no obligation to take any AMD products. What's more, it was also ruled that AMD should have known that they weren't going to get the 386, because Intel didn't take any AMD products. Yet AMD made no attempt whatsoever to get a 32-bit processor on the market in alternative ways (for example, by reverse-engineering the 386 right away). And AMD also did not seek legal action. Therefore, AMD was also in the wrong, by basically just waiting for Intel to hand over their IP, and not doing anything about it until years later. Which means that AMD is basically responsible for Intel managing to build the 386 monopoly.

An arbitrator then basically ruled: "Well Intel, what you did was not in good faith, so give AMD the 386 rights anyway, even though AMD did nothing to earn these rights".
Intel appealed, but the court upheld the arbitrator ruling. So AMD got the 386 rights in the end, but not because AMD actually was entitled to them by their original license. In fact, the arbitrator argued that because Intel did not negotiate in good faith, the original agreement was breached. And this breach of agreement had hurt AMD, so they would have to be compensated for their damages.
All this is very... arbitrary. Point remains, that it required an alleged breach on Intel's side of the agreement for AMD to be awarded the rights to the 386 (while giving nothing in return).
In other words: had the agreement not been considered breached, AMD would not have had to be compensated, and would not have received the rights to the 386, because the agreement never entitled AMD to them, and AMD knew.

The only claim that AMD actually won is that the number '386' cannot be used as a trademark. Which explains why Intel moved to a different naming scheme with the Pentium (486 was already out by the time of this court ruling).

mpe wrote:

AMD/other vendors opinion was obviously very different.

And their opinions are irrelevant since Intel was owner of all IP of the x86 architecture and most implementations of it, including the 386. That is a fact, a fact not disputed by any court. Hence the licensing.
AMD and others basically wanted to freeload on Intel's success. This is exactly why patents and other IP laws exist: to protect the investment of the inventors/creators.

Yes. They had some sort of open ended partnership where AMD will continue to have access to Intel’s product line (for hefty royalties) if Intel accepted some other AMD products in return. At one point Intel though it is not really working for them and deliberately stopped accepting AMD products (even these that would be valuable for them) in order to avoid them getting access to the 386 and frustrate competition. This was found unfair by the arbiter and the court.

AMD wasn’t the only one. Intel was also suing C&T, NEC, Cyrix, UMC and possibly others. Some withstood others didn’t.

But whilst the legal process is interesting. It is not that important other than it delayed the Am386 for some time. In the end the AMD was awarded the license. Since they got the license the Am386 cannot be seen as illegal or unlicensed chip (reverse engineering isn't illegal). It went to the market, later than planned (a kind of Intel's win) and was hugely successful.

mpe wrote:

At one point Idntel though it is not really working for them and deliberately stopped accepting AMD products (even these that would be valuable for them) in order to avoid them getting access to the 386 and frustrate competition.

Which is perfectly fair by the terms of the license.
I mean, AMD argued about some matrix display controller that Intel was not interested in. Apparently that was the best that AMD could offer.
How does that compare to a 386 CPU?
Clearly AMD's offering was not even remotely comparable to Intel's offerings. 'Valuable' is a very relative term. Heck, basically the whole AMD company floated on manufacturing chips licensed from Intel.
In fact, if you know your history, you know that the whole second-source licensing for x86 CPUs was mainly forced on Intel by IBM. So it was never something that Intel wanted to do in the first place. Because it would never work out for them.

mpe wrote:

This was found unfair by the arbiter and the court.

As I say, that's very arbitrary. It's a strange situation when your 'competitors' rely on you to supply the actual chip designs and patent rights. That's not competition to me.
Clearly that situation was never going to work for Intel, as Intel was the only one actively developing the x86 architecture. Others were basically freeloading on Intel's tech, as a side-effect of IBM wanting second-source.

If Intel wasn't in the position where they could easily control the x86 market, and other competitors could actually offer meaningful IP in return, the situation would have been entirely different.
But now everyone was just taking from Intel and giving nothing meaningful in return. Is that fair? I don't think it is.

mpe wrote:

It is not that important other than it delayed the Am386 for some time.

I think it is when people are actually claiming that the Am386 was not an illegal copy. Let alone that Intel lost the claim to its legality.
It very much was, which is why Intel could delay it.
It was made legal by arbitration. It wasn't legal in and of itself. The cross-license agreement did not give them the rights to anything beyond 286.

And the impressive 386DX-40 still got sold by the bucket loads. To be continued......

Scali wrote:

[quote.
It was made legal by arbitration. It wasn't legal in and of itself. The cross-license agreement did not give them the rights to anything beyond 286.

Intel would have had to agree to he out come of the arbitration. That's the whole point, all side agreed, the matter was settled, aka the out come was legal in the courts. Whether you agree or disagree it doesn't really matter. Every one else moved on.

Caluser2000 wrote:

Intel would have had to agree by the out come of the arbitration Aka it's was legal.

Well no. Arbitration, like a settlement, can contain agreements that in and of themselves may not be legal (because sometimes that is the only way to end a dispute). They are however binding.
As I say, the arbitration *made* the 386 legal, because the arbiter forced Intel to license all relevant IP to AMD royalty-free.
Prior to that decision of the arbiter, the 386 was in fact illegal, because AMD did not have the rights to the relevant IP (so Intel was right).
The arbiter basically null-and-voided the original second-source agreement, and set up a new agreement, under which the 386 would then be legal.
So it was not always legal, it was made legal, because AMD basically got the rights for free in arbitration. This is in fact against patent regulation. The holder of a patent owns the IP, and is normally free to decide to whom they may or may not license it, and what kind of compensation they want in return.

Ergo, prior to AMD obtaining the required rights, it was in fact, illegal (as per Intel's second-source agreement). It would have been different if someone had ruled that the second-source agreement did in fact also include the 386, but they did not. Because then AMD would have had the rights by default. Or it would have been different if the patents themselves had been ruled invalid, so that AMD did not need to acquire any rights. Again, they did not.
Ergo, the 386 was not legal prior to the new patent deal resulting from the arbitration. AMD was in violation of Intel's IP. The new patent deal fixed that situation.

So basically, it was AMD who was in the wrong with trying to sell the 386. Intel was not in the wrong by not giving them those rights.
The decision of the arbiter does not really reflect that, because it looks like AMD won. Intel was only in the wrong for vague reasons like "monopoly" and "not in good faith".

I bet you didn't sleep last night either 🤣. Once the final decission was made everyone everyone moved on and the wonderful 386DX-40 was produced and Cyrix dips it's toes without interference from Intel thank goodness. But there's more.......

Caluser2000 wrote:

And the impressive 386DX-40 still got sold by the bucket loads. To be continued......

Couldn't the Intel 386-33 just have been easily overclocked to 40? I have an SBC with an AMD 386DX-40 and an Intel 387 (16-33) and pretty sure it is clocked at 40Mhz.

It was just that Intel didn't officially release a 40Mhz part.

cyclone3d wrote:

Caluser2000 wrote:

And the impressive 386DX-40 still got sold by the bucket loads. To be continued......

Couldn't the Intel 386-33 just have been easily overclocked to 40? I have an SBC with an AMD 386DX-40 and an Intel 387 (16-33) and pretty sure it is clocked at 40Mhz.

It was just that Intel didn't officially release a 40Mhz part.

Intel just stopped at 33Mhz as the 486 was in production. Of course they could have reduced the the price and sold more units but had introduced the cost cutting 386SX to cover the lower end 386 market and cut out the high end 286s. But that wasn't the case. In reality 386SX and 286 with the same clock speed the 286 was the same if not a bit more efficient carrying out 16-bit instructions which most software of the day was using. With the AMD386DX-40 it clobbered those 386SXs in one grand swipe. Computing Mags at the time were on it like white on rice and of course the legal rangle was just added publicty.