What came after 486?

CPUs didn’t have brand names, besides the manufacturer, until the 1990s. They had part numbers and clock speeds. Frequently we shortened the part numbers. The 486’s full part number was 80486. The courts wouldn’t let Intel trademark a number, so the 486 was the last CPU of its kind, raising the question: What came after 486?

The follow-up for the 486 was the Pentium, at least in Intel’s case, and it was introduced March 22, 1993. But several companies made 486 CPUs, and several of those released their own follow-ups to the 486, including AMD and Cyrix.

Intel’s Pentium CPUs

The original Pentium CPUs of the 1990s used a faster front side bus than the 486 and plugged into a different socket. Intel did sell a Pentium Overdrive CPU that plugged into a special 486 socket with an extra row of pins. This hybrid had a Pentium core, but still used the slower 486 front side bus so its performance was somewhat disappointing.

A true Pentium was about 40% faster than a 486 running at the same clock speed, and the first Pentiums ran at 60 and 66 MHz while the fastest 486 at the time also ran at 66 MHz. Intel eventually released 486 CPUs clocked at 100 MHz, but by that time, it had ramped Pentium clock speeds up to 120 and 133 MHz. The original Pentium-branded CPU topped out at 200 MHz in the end, and the Pentium MMX-branded CPU reached 233 MHz.

For about three years, Intel had this post-486 market all to itself. Had this not been the case, the Pentium FDIV bug would have been a disaster for Intel.

Why brand name was important after the 486

Intel had mixed success in stopping companies from making clones of its chips, and no success whatsoever in keeping them from using Intel’s numbering scheme to indicate their chips were functionally equivalent. That meant Intel never produced a chip it called the 80586, which would have been the logical followup to the immensely successful 80486. Intel codenamed its 486 successor the P5, and when it released it, Intel gave it the brand name of Pentium. The name invoked the number five, but was completely trademarkable, unlike the number 586.

Other companies also intended to produce 586-class CPUs. Intel’s profit margins were high enough they figured they could undercut Intel’s prices a little and still turn a reasonable profit. And for a while, they followed that naming scheme, but that wouldn’t work forever. Intel knew it, and they knew it. The result was market confusion, some of which persists to this day.

But we’re probably getting ahead of ourselves a bit here. Motorola was the other major manufacturer of CPUs at the time, and Motorola didn’t have the problem of other companies making knockoffs of its CPUs. Why was Intel’s situation different?

Why so many companies made 486s and Intel wanted them to stop

At least seven companies made 486-class CPUs and called them by that name. This dates back to IBM’s original requirements for the first IBM PC. IBM wanted multiple sources, in case Intel ever couldn’t meet demand for any reason. So Intel licensed production of the 8088 and, later, the 8086 and 80286 CPUs to multiple companies to give IBM the insurance it demanded. Eventually, even IBM itself licensed the rights to produce Intel CPUs.

Things changed with the 386. Initially IBM didn’t want to produce PCs based on the 386, because they didn’t PCs to be that powerful. If you wanted 386-type power, IBM wanted to sell you a System/38 minicomputer, the predecessor to the AS/400 or what IBM today calls its System i series.

With no contract from IBM requiring it to do so, Intel didn’t license the 386 to all comers. Intel figured it had enough production capacity to meet demand, and no one else had the clout to demand Intel provide a second source.

AMD’s 386 and 486 clones

One longtime Intel second source, AMD, decided to push the limit and become a competitor. Intel didn’t provide AMD everything they needed to produce a 386, but they used what they knew from making earlier chips under license and reverse-engineered the rest, and released a very close clone of both the 386SX and 386DX, without Intel’s blessing. Intel sued and AMD counter-sued. After five years of litigation, the courts ruled AMD could produce its 386 CPUs.

Then AMD did the same thing with the 486.

AMD didn’t do quite as well with the 486. AMD’s 486 was also a close clone, but Intel won some concessions regarding AMD’s higher-end 486s. AMD could use the instruction set and the pin connector, but somehow Intel convinced the courts to force AMD to run the chips at a lower voltage. This protected Intel’s lucrative upgrade market. Still, Intel had to share the 486 market with AMD, and while most top-tier PC brands stayed with Intel, Compaq and Acer gladly used the AMD chips in some of their systems. AMD even produced a 66 MHz 486SX2 at Compaq’s request.

And AMD did very well in the generic PC market. Motherboard makers just put jumpers on their boards so you could select voltages appropriate for AMD or Intel.

But the 486 was the end of the line of this approach for AMD. If AMD wanted to make CPUs beyond the 486 era, it was going to have to come up with its own designs.

UMC’s 486 clones

Intel was able to sue UMC to stop it from producing its 486 clone. UMC was a maker of motherboard chipsets and wanted to expand into the CPU business as well, pretty much the reverse of what Intel did during the same time period. UMC’s 486 violated one of Intel’s memory management patents and UMC didn’t have a way to get around it.

IBM 386 and 486 derivatives and clones

IBM produced chips based on both Intel’s 386 and 486 designs, but used them in its own systems and paid Intel a royalty to do so under an option IBM exercised in April 1989.

The agreement didn’t allow IBM to sell its CPUs separately, though it could sell them as part of a motherboard. In a bizarre twist, IBM’s 66 MHz 486SLC2 became a popular clone motherboard in generic PCs for a while. But the IBM CPUs based on Intel technology always had some catch that made Intel’s version more desirable, which put IBM in the unfamiliar situation of being an off-brand and having to sell at an off-brand price.

IBM could and did produce exact copies of the Intel CPUs. These chips still had Intel branding with an indication that IBM manufactured it. Intel required IBM to buy a quantity of Intel-manufactured CPUs for every CPU it manufactured itself, so many IBM PCs from this era have regular Intel CPUs in them.

Not every IBM 486 was Intel-derived, however. IBM also produced chips for Cyrix, and as part of that arrangement, IBM got part of the chip yield. IBM generally chose to sell those chips on the open market rather than use them in its own PCs.

Cyrix 486 clones

Unlike AMD, Cyrix’s 486 was its own design. The first Cyrix 486s were really 386/486 hybrids, but later Cyrix 486SX and DX CPUs were pin-compatible with Intel. They weren’t quite as fast as an AMD or Intel of the same speed, but were generally within 10 percent. Cyrix got around the patent issues that stopped UMC by having companies who had cross-licensing agreements with Intel manufacture its chips. When you see a 486 with TI, SGS-Thompson (ST Microelectronics), or IBM’s brand name and logo on it, it’s a Cyrix design. Cyrix 486s also exist with its own brand name.

Cyrix and its partners pretty much had to live with the low-tier clone market. They didn’t attract any big names like Compaq.

Non-Intel Pentium successors to the 486

Both AMD and Cyrix produced chips they called the 5×86. The AMD 5×86 was nothing more than a souped-up 486, with more cache and a higher clock multiplier. It plugged into 486 motherboards and had a slower memory bus. It was fast enough to provide Pentium-class performance, but an AMD 5×86 running at 133 MHz ran more like a Pentium running at 90 MHz.

Cyrix’s 5×86 was a hybrid design. It used a 486 bus and plugged into the 486 socket, but actually incorporated some Pentium-like technology into its CPU core. The 486 bus slowed this chip down as well, but it outperformed AMD running at the same clock rate.

True Pentium-class chips

A startup called NexGen produced a chip it called the Nx586. This chip used a different socket and chipsets but its performance, at least when running non-math-intensive tasks, was comparable to Intel. The Nx586 had no math coprocessor, however, so it was really the equivalent of a super-fast 486SX. Compaq invested in NexGen and left the door open to using its chips, but the lack of a math coprocessor was a showstopper for a first-tier PC vendor.

AMD and Cyrix released true Pentium-class CPUs in 1996 that plugged into a Pentium socket and worked with a Pentium chipset. AMD’s chip was called the K5, while Cyrix called its Pentium-class chip the 6×86. Claiming sixth-generation performance ended up being problematic for Cyrix, because it was only faster than the Pentium in integer operations. Pentiums were much faster at math that involved decimal points, which included 3D gaming. With 3D gaming becoming more popular, Cyrix faded. A small number of SGS-Thompson-branded 6x86s exist, but IBM made most 6×86 chips and sold its share of the chips as IBM 6x86s. Functionally they are identical to their Cyrix counterparts.

AMD’s K5 was very late to market and AMD couldn’t get its speeds above 133 MHz. AMD bought NexGen in 1995 and NexGen’s design team was responsible for AMD’s very successful K6.

IDT, a maker of memory chips, also entered the market with a Pentium clone it called the C6 and/or Winchip. Like AMD and Cyrix, IDT couldn’t match Intel’s math performance. The C6 ended up being a budget chip and eventually VIA bought the rights to both it and Cyrix’s CPUs. The IDT design lived on in VIA’s x86 designs.

Another very short-lived Pentium clone was the Rise MP6. It never gained much traction in the 90s but its core lives on in the Vortex86.

AMD going it alone against Intel

After the Pentium II, Transmeta tried to compete using its Crusoe CPU, and VIA released a Pentium III-compatible CPU called the C3.

But only AMD managed to continue releasing truly viable competitors to Intel. The technical and legal challenges just proved to be too great. While the Pentium-class CPUs used the same processor bus and chipsets, with the Athlon, AMD switched to a bus it licensed from Digital Equipment Corporation. AMD’s chips require a different chipset and therefore a different motherboard than Intel, so it’s not a true two-source situation, but most PC makers now use chips from both suppliers, picking and choosing components to hit specific price points. Some companies, notably Dell, remained Intel-only well into the 21st century, but eventually even Dell found it was more beneficial to mix and match rather than use Intel exclusively.

The branding can still be confusing, but since Intel settled in to using its Core i3, i5, i7, and i9 nomenclature, AMD followed with its Ryzen 3, 5,7, and 9 branding and numbering scheme to indicate which class of CPU performance it targets.

It’s just as hard today to keep various chip generations’ relative performance straight as it was in the 486 days, but at least the information is easier to come by today.