Last Updated on March 2, 2026 by Dave Farquhar
UMC wasn’t the only company with a forbidden CPU in the 90s. Fellow chip set maker Chips and Technologies took a somewhat similar approach late in the previous generation, with an independently created clean room implementation of the 386. This short-lived CPU may be the rarest 386.
Faster than Intel
Like the UMC 486, the Super 386 outperformed the equivalent part from Intel or AMD. The difference wasn’t quite as large as UMC, but a 10% boost is nothing to sneeze at, especially if they took it a step further and delivered higher performance at a lower price. The Super 386 existed in 25, 33, and 40 MHz versions.
The Super 386 implemented a second pipeline, which was the main reason it was faster than the equivalent AMD or Intel 386. This was one of the tricks the 486 employed to be faster than the 386. There was a second version of the Super 386 that was not pin compatible with the 132 pin Intel 386, but rather used its own 144 pin socket. This version implemented a small onboard level 1 cache, which was another speed trick the 486 used. It was a much smaller 512 byte cache than the 8K most Intel 486s had, but any high speed level 1 cache is better than none at all. This version of the Super 386 really would have straddled the line between the 386 and 486.
But this is a rare and infrequently documented CPU, which is why you may have never heard of it. If you’ve never seen one, that’s because it wasn’t on the market very long.
Avoiding legal problems when cloning a CPU
To do a clean room implementation, you hire one or more engineers who have the aptitude to carry out a design, and who have done similar work in the past, but can demonstrate that they have never seen or worked with the particular design they are supposed to clone. An engineer with chipset experience but no CPU design experience might be okay, and an engineer with CPU experience but nothing Intel compatible would be ideal. Another set of engineers looks at the publicly available documentation, then write their own specifications based on that documentation for the design team to implement.
Implementing in this way shields you from copyright infringement. Because of the differing backgrounds of the designers, the two designs will have significant differences. There may be some incidental similarities, but if all the designers sign affidavits ahead of time stating that they never saw the other design, is probably not difficult to convince a jury that there was no other logical way to design that part of the circuit.
The design will look very different from, say, VTI’s implementation of the Atari TIA chip, which looked like it was simply a copy of the Atari chip, sliced into quadrants, and then rearranged.
Intel’s legal action against the Chips and Technologies Super 386
But what this method doesn’t protect you again is patent infringement. Intel sued, claiming that Chips and Technologies violated five of their patents, three more than they accused UMC of infringing.
Chips and Technologies had an ace up its sleeve that UMC did not. Chips and Technologies outsourced production to Texas Instruments, kind of like Cyrix. Texas Instruments had licenses to all of the relevant patents, so in all likelihood, Chips and Technologies would have been able to fight off the lawsuit and emerged victorious. The same defense worked for Cyrix. The trouble was, Chips and Technologies didn’t have the capital to survive a prolonged legal battle. So instead, they withdrew the Super 386 from the market.
This is one reason why there is no Super 386 SX. They released the DX version first, and the SX version never went into production. The 144 pin version that would have been significantly faster also does not appear to have reached mass production.
What was interesting about the Chips and Technologies Super 386
The Super 386 is interesting because it would have been an even better competitor to the low end 486 than AMD had been. But the problem was that the version you really would have wanted that included level 1 cache required a different and incompatible 144-pin socket. Cyrix had a 132 pin CPU that plugged into a 386 socket called the 486 DLC. It included a small level one cache. Cyrix had found a way to pack a level 1 cache into the existing design without needing additional pins. It just needed a few changes to the BIOS to enable the functionality. Changing the BIOS is much easier than designing in a second socket. The Cyrix solution was more cost-effective.
Arriving as late in the game as it did, I don’t know how much impact the Super 386 would have had. It’s unlikely it would have affected Intel market share very much, and its market share would have come at the expense of AMD and Cyrix instead.
What’s potentially more interesting is what Chips and Technologies would have done if they had been able to survive into the 486 era. Would they have created a straightforward implementation like Cyrix did, or would they have found a few more performance enhancing tricks like UMC did? We will never know.
Chips and Technologies’ eventual fate
After losing the court battle over the Super 386, Chips and Technologies retreated to building graphics chips. And in 1997, their former nemesis Intel acquired them to get their graphics technology. That closed the book on Chips and Technologies, the inventor of the PC chipset as we know it, and the would be rival CPU designer.
The lawsuit did receive a reasonable amount of press coverage, at least in the technical press. If you were reading Infoworld in 1991 and 1992, you were absolutely aware of the Super 386. But unlike the UMC CPU, I do not recall ever seeing a reference to the Super 386 and any motherboard manuals when I was building 386 systems. I didn’t build as many 386 PCs as I did 486, so that may be why. Or maybe the motherboard manufacturers had deleted any reference to the Super 386 by the time I was building.
David Farquhar is a computer security professional, entrepreneur, and author. He has written professionally about computers since 1991, so he was writing about retro computers when they were still new. He has been working in IT professionally since 1994 and has specialized in vulnerability management since 2013. He holds Security+ and CISSP certifications. Today he blogs five times a week, mostly about retro computers and retro gaming covering the time period from 1975 to 2000.