Miphee wrote on 2021-01-11, 13:56:

It couldn't be imported either, why?

Because UMC knowingly violated Intel patents. They were sued by Intel, and UMC countersued challenging the validity of the patents. UMC lost and settled with the CPU being forbidden from being sold in the US, them paying damages and dropping the challenge to the patents.

http://www.cpushack.com/2012/09/06/intel-vs-t … the-338-patent/

Thanks. But what was the point of building a CPU that knowingly infringed on a huge corporation's copyrights?
Did they think that Intel won't notice and they can get away with it?
Can't really see the logic behind it. Did UMC sell this CPU in the US before the lawsuit?

Miphee wrote on 2021-01-11, 16:13:

Thanks. But what was the point of building a CPU that knowingly infringed on a huge corporation's copyrights?
Did they think that Intel won't notice and they can get away with it?
Can't really see the logic behind it. Did UMC sell this CPU in the US before the lawsuit?

I believe so for a short while.

The rest of the planet didn't care about the US law suite though.

Miphee wrote on 2021-01-11, 16:13:

Thanks. But what was the point of building a CPU that knowingly infringed on a huge corporation's copyrights?
Did they think that Intel won't notice and they can get away with it?
Can't really see the logic behind it. Did UMC sell this CPU in the US before the lawsuit?

UMC was taking a calculated risk. They probably guessed they would get into legal trouble in the US, but that was never their intended market for this CPU - it was a relative latecomer and aimed at low-cost computers for developing markets. They figured they could undercut Intel, AMD and Cyrix there and still turn a profit. No idea whether they eventually did, but not being able to sell a CPU in the US that would hardly sell there anyway (this CPU was released in 1993, but only sold in volume in 1994, by which time a DX2/50 was low-end in US/EU) was hardly an obstacle for the unicorns.

Miphee wrote on 2021-01-11, 16:13:

Thanks. But what was the point of building a CPU that knowingly infringed on a huge corporation's copyrights?
Did they think that Intel won't notice and they can get away with it?
Can't really see the logic behind it. Did UMC sell this CPU in the US before the lawsuit?

No, it was never sold in the US. The issue here is that back in the 70s before Intel became Chipzilla they had cross-licensing agreements with half the industry - TI, AMD and others who fabbed their chips to increase market share. As Intel grew in size they also got...arrogant, and due to some rather lenient policies in the USPTO, they were able to patent the entire memory segmentation/page table scheme on the i386 (see USPTO 4972338A, aka the 338 patent). Intel used the 338 and some other patents as a legal hammer to beat down competitors, or at least used it to tie them up in court. If you want make anything 386+ compatible you’ll need to pay Intel for a license to use it (assuming that they want to sell) or violate that patent. In truth, all of the old-school x86 chip makers spent time in court because of the '338, but for different reasons. Here's a summation:

AMD - Team green and team blue had a cross-licensing agreement back in the 80s where Intel was supposed to give AMD access to their Microcode (since they fabbed x86 CPUs and far as AMD are aware, Intel is supposed to supply any and all CPU microcode until the agreement runs out, while Intel thought that they agreement only covered 8086 and 80286, and not every subsequent CPU afterwards). After Intel cut AMD off for anything beyond the 286, AMD sued Intel for access to the 386/486 microcodes while also bundling copies of Intel’s microcode (which contained the 338), assuming that their interpretation of the agreement is correct - that aspect is tied up in litigation from '85 onwards. Intel counter-sued AMD for violating 338 on their AM386 and 486 chips. The entire thing bought new yachts for the lawyers on opposing sides until a settlement was reached in ‘95. Intel gave AMD access to the 386/486 microcodes and a license to use the ‘338, while AMD settles the lawsuit for the breach of agreement and writes Chipzilla a 40 million dollar check. Since the Pentium was designed after the agreement was supposed to run out, well, AMD didn't get access to that.

Cyrix - well, Cyrix is a little odd. They didn’t have a licensing agreement but they implemented the 386/486 microcode via reverse engineering and came up with a clean room implementation that did not use any Intel microcode, and as far as Intel lawyers were able to determine in a court case, it didn’t technically violate patent 338 (it's a workalike but didn't work the same way as the patent). And since their fabber is TI (which is an Intel cross-licensing partner), SGS-Thompson (same) or IBM (yep) , they didn’t really get in trouble. Cyrix went ahead and sued Intel for using 2 of their patents, one for register renaming and one for power management...both used on the Pentium Pro. Intel and Cyrix settled out of court but Cyrix got some money and a fairly wide-ranging cross-licensing agreement, while Intel used NatSemi fab techniques on the P6. The legal costs did run up for Cyrix and they eventually had to sell out to NatSemi.

UMC - that’s the headscratcher - they were not a licensee nor are they an original Intel chip fabber - they had their own fabs and had a clean room design, but they did not have a ‘338 licensing agreement. They seemed to have forgotten about the earlier, more incestuous relationships between Intel and its various fabber fabs. Intel sued them in US court while UMC challenged the validity of the 338 - and they eventually settled. UMC will stop making 338 violating chips and stop questioning 338 while Intel will leave them alone. The fact that they were spun off from the Taiwanese government Industrial Technology Research Institute (much like Mediatek and TSMC) probably gave them some backers high in government and kept them out of too much trouble, and Intel probably didn't want to piss off the Taiwanese too much with the low orbit legal cannon.

Miphee wrote on 2021-01-11, 16:13:

Thanks. But what was the point of building a CPU that knowingly infringed on a huge corporation's copyrights?
Did they think that Intel won't notice and they can get away with it?
Can't really see the logic behind it. Did UMC sell this CPU in the US before the lawsuit?

I'm no lawyer, but..

Patents are like weapons, even though my old physics teacher stated otherwise (basically saying patents were there to protect intellectual property of the inventors).

In essence, it doesn't matter who wins at the court.
Filing a case is useful for slowing down the opponent until his product is somewhat outdated.

So, - oversimplificated stated -, it doesn't matter if a CPU was legally reverse-engineered or not.
When the truth comes out, after a long battle at the court, it's too late, anyway.

So why not just selling a pirate copy straight from the start ? 😉

Edit: From my point of view, as a foreigner, the US laws seem strange, anyway.
If I got it right, then the US laws/rules force a company to be aggressive and mean.
A company has to sue a person or company if the trademark or patent was violated.
If it does not, the patent/trademark is rated as "not worthy". Simply said. Thus, the patent/trademark in question might expire.
That being said, I'm speaking under correction here.

Because Intel.

Also, Fuck Intel.

I got all the answers I needed and then some! It was very informative, thank you all!
To be honest if these grey areas wouldn't exist Intel would probably dominate the entire chip market and that would certainly violate antitrust laws.
Thank god for smaller chip makers!