You really should brush up on your CPU history, Coug. Try Tannin's site. :wink: He's a bit up himself, of course, but he knows his stuff in that department.
R&D was the whole idea behind the Cyrix/IBM relationship. At that time, IBM had their own CPU design team. Well, two of them, actually: the X86 design team and also the PowerPC guys. Plus the guys who did their big iron, no doubt, unless that was handled by the PowerPC team - I don't follow that end of the market. IBM had a series of in-house designed X86 CPUs, which were made using a combination of Intel-licenced technologies and their own in-house enhancements. The best known and most successful of them were the 486SLC and the 486BL, or Blue Lightning.
The Blue Lightning was essentially an Intel 386DX plus IBM's own developments. The one in the picture is a 75MHz one and it was more than capable of holding its own against the Intel 486 parts.
Now, about the R&D. Cyrix were fabless. They had no manufacturing facilities at all. They used to rent fab facilities from people like Texas Instruments (Jerry was an ex-TI guy in the first place, remember) or SGS-Thompson (who now have some other name which I forget, but are a
huge multinational firm of French origin who make everything from microchips to military radars and missile parts).
IBM had plenty of production capacity but a shortage of design engineers, Cyrix had a world-best team of designers but no fab plant of their own. There were complications and fine print, but essentialy the deal was simple. Cyrix would do all the design work (allowing IBM to redeploy their X86 guys over to the PowerPC) and IBM would take care of the manufacturing. Then they would each take 50% of the finished chips.
It worked very well for some years. They sold a
hell of a lot of chips. But eventually management changes at Cyrix saw them try to become more sales-focussed and less product-focussed and their design team - up until that time unquestionably the best X86 design team in the world - lost the plot. Or were hamstrung. Or couldn't get enough development and testing wafer space out of IBM. Or lost all their best talent - the stories seem to vary. The net result, though, was simplicity itself: they had no new product worth talking about, and in the CPU game, that is a death warrant.
In short, IBM never had anything to do with the R&D of the Cyrix/IBM chips inb the first place. So long as Cyrix remained a hard driven, product-focused design house, they could (and did) match it with all comers. From the moment they went all touchy-feely and decided that cuddling up to the buying department at Compaq was more important than designing great CPUs, their death warrant was made out, and by failing to concentrate on their design strength - fast CPUs for the desktop - and pissing about with fifty different schemes for set top box chips instead, it was soon signed.
But let us not forget: in their golden era, they were a truly formidable competitor. AMD had terrible trouble trying to compete with both Intel and Cyrix at the same time - they were really stuck in the middle: not as respectable as Intel and not as cheap and fast as Cyrix. Thanks largely to their flash RAM division, they were able to stick it out for long enough for their newly-accquired NexGen designers to come up with a competitive chip - the K6 - and with this great design they managed to survive and compete for just long enough to finally get rid of their very troublesome 0.35 micron process and move to their 0.25 micron one, which was an immediate winner. From that point on, AMD were looking good.
Cyrix were the company, far more than any other, that made computing affordable. And yes, I include AMD in that statement.
PS: I remembered SGS-Thompson's new name: ST Microelectronics. At least that's the name of one of their divisions. They are a bit like IBM or NEC or Samsung - make all kinds of stuff.
