Cyrix

CPU manufacturer Cyrix began in 1988 as a specialist manufacturer of high-performance maths co-processors for 286 and 386 systems. The company was founded by ex-Texas Instruments staff, and had a long but rather troubled relationship with TI from that time on.

Cyrix founder Jerry Rogers was a hard-driving leader who head-hunted the best engineers in the USA and pushed them mercilessly. His tiny 30-man design team set themselves the task of first matching, and then beating performance leader Intel at their own game. Much of the credit for this belongs to Jerry Rogers himself. As an aggressive, go-getting leader he took Cyrix from nothing to the mighty 6x86 - a faster chip than Intel's best.

But it was no use having the best chip in the world if few were buying it. Cyrix had great product and fantastic pricing but needed a major OEM customer - an IBM, a Compaq, or a Dell to give them long-term, steady contracts. They had always done well in the independent market with the smaller specialist shops, but a well-balanced order book has a mix of customers, large and small. It was not impossible for a non-Intel manfacturer to sell CPUs to the major PC makers - AMD had done it after all, - but Cyrix could never quite close the big deals. Eventually they dispensed with Jerry Rogers as CEO and added marketing people to the board. Within just a few months they had sold the MediaGX to Compaq and other big sales soon followed.

Cyrix had always been a fabless manufacturer: in other words, they designed and sold their own chips, but farmed the actual manufacture out to outside contractors. In the early days they mostly used Texas Instruments production facilities, and they also used SGS Thompson extensively. In 1994, following a series of disagreements with TI, and in search of high-tech, thin-wafer production to match it with AMD and Intel, Cyrix turned to IBM Microelectronics. For the next four years, Cyrix CPUs were all made by IBM. As part of the deal, IBM got to make their own Cyrix-designed CPUs at a very reasonable royalty.

Unlike AMD, Cyrix had never manufactured or sold Intel designs under license, but designed everything in-house, from the ground up. So while the AMD's 386s and even 486s had some Intel-written microcode software, the Cyrix chips were all completely independent, clean-room designs.

Intel spent many years in the courts, claiming that the Cyrix 486 violated Intel's patents. (Just as they did with every other x86 CPU manufacturer right up until 1998.) By and large, Intel lost the Cyrix case. But the final settlement was out of court: Intel agreed that Cyrix had the right to produce their own x86 designs in any foundry that happened to already hold an Intel license. Both firms gained out of this: Cyrix could carry on having their CPUs made by Texas Instruments, SGS Thompson, or IBM (as it happened, all holders of Intel cross-licenses); Intel avoided complete open slather in the x86 market. Both firms were (presumably) happy to avoid yet another round of mindless and expensive litigation. The follow-on 1997 Cyrix-Intel litigation was the reverse: instead of Intel claiming that Cyrix 486 chips violated their patents, now Cyrix claimed that Intel's Pentium Pro and Pentium II violated Cyrix patents - in particular, power management and register renaming techniques. The case was expected to drag on for years but eventually be settled out of court. In fact it was settled quite promptly, by another mutual cross license agreement. Intel and Cyrix now had full and free access to each others patents - a remarkably sensible resolution that let both firms get on with making better chips. Notice that it neatly avoided the issue of law: such a settlement didn't say whether the Pentium Pro violated Cyrix patents or not, it simply allowed Intel to carry on making them either way - exactly as the previous settlement side-stepped Intel's claim that the Cyrix 486 violated Intel patents.

In August 1997, while the litigation was still in progress, Cyrix merged with National Semiconductor (who already held an Intel cross-license). This provided Cyrix with an extra marketing arm and access to National Semiconductor fab plants, which were originally constructed to produce RAM and high-speed telecommunications equipment. (So far as manufacturing technology goes, RAM and CPUs are very similar.) The IBM manufacturing agreement remained for a while longer, but Cyrix eventually switched all their production over to National's plant. The merger improved Cyrix's financial base, and give them much better access to development facilities.

The merger also resulted in a change of emphasis: National Semiconductor's priority was single-chip budget devices like the MediaGX, rather than the higher performance chips that Cyrix had traditionally made. They clearly lacked confidence in Cyrix's ability to compete at the high end of the performance spectrum - a rather odd decision, as Cyrix had long since proved their ability to come up with outstanding CPUs. Once Cyrix had National Semiconductor's fab plants and reserves of capital behind them, they should have been more able to compete with AMD and Intel's higher-end products, not less. Perhaps National's reasoning was that the high-end looked promising for Cyrix but the low-end MediaGX family parts had no real competition and were an absolute certainty.

By mid-1999, it was clear to all that Cyrix were in trouble: it took them almost a year to push the MII from PR-300 to PR-333, and despite occasional promises, the flow of good new products was no more than exactly that - promises. Cyrix parts gradually fell further and further away from the performance mark, and their natural application shrank: to the mid-range, to the entry-level, and eventually to nothing at all. As things stood, it was obvious that Cyrix would soon disappear forever. Cyrix was left to stagger along for several months longer: bereft of direction, with morale at an all-time low, and losing their best engineers one by one. In the end, VIA bought the feebly-twitching remains. By this time the original development team was no more, the products were hopelessly out of date, and the only thing left of value was the Cyrix name.

Over their short history, Cyrix had a huge influence on the average PC buyer, even the buyer who never owned a Cyrix chip. Between them, Cyrix and AMD lowered the cost of computing by many hundreds of dollars - and lowered it for both their own customers and for Intel buyers too. Every time a consumer bought a PC with an Intel chip, she saved several hundred dollars simply because Cyrix (and AMD) had forced the market leader to moderate its prices. Every time a consumer bought a PC with a Cyrix chip, she saved even more. As a salable product line, Cyrix are gone beyond all hope of recovery now, but their legacy of affordable performance remains to this day.

This is an edited version of a document that first appeared at http://www.redhill.net.au, and is used with permission. Some of the more unsuitable material has been omitted but it still needs rewording to fit the Wikipedia style.