JC over één jaar Athlon

JC heeft een interessant artikel geschreven over de gebeurtenissen (goed/slecht) die er sinds de gedenkwaardige AMD dinner meeting van 10 juni 1999 plaats hebben gevonden. Tijdens deze meeting hield AMD engineer Dirk Meyer een presentatie waarin de architectuur van de K7 voor het eerst openbaar werd gemaakt. Dirk Meyer was eerder werkzaam voor Digital, waar hij aan de wieg stond van de Alpha 21064 en later meewerkte met de ontwikkeling van de nog steeds zeer gerespecteerde Alpha 21264.

Na de dinner meeting volgde op 8 augustus de officiële release van de Athlon. Wat er sindsdien is gebeurd zal velen nog vers in het geheugen liggen en heeft AMD gebracht tot de meest succesvolle periode in haar bestaan:

15 GOOD, 7 (and a half, if you count that mini-one) BAD. Not a bad ratio, and the numbers make the situation worse than it actually is. Most of the BAD issues are either minor ones or are in the process of being (or have already been) canceled out by a GOOD. AMD is positioned as the leader of their primary industry from a technical perspective, with their only true disadvantage being having only one full Fab and one partial Fab, while the competition has many such plants. Lack of compiler support was a critical issue, but this is being alleviated.

Even with most of the GOODs discarded, AMD is guaranteed an incredibly profitable and fruitful year 2000. There is only one uncertainty that could cause this health to falter: Willamette. Intel's upcoming competition for the Athlon, Willamette promises little in the way of added performance-per-clock, but does seem to have the potential for incredibly high frequency ramping. If AMD can more or less keep up to Willamette MHz, they will likely remain profitable and healthy for several years more. On the other hand, if Intel's processor starts gapping above AMD's maximum offering, then AMD's profits will slowly dwindle. Because of their FLASH memory business explosion (another GOOD, but not included in the assessment, because it has little to do with the Dirk Dinner), AMD has a good chance of sustaining profit in even the worst case scenarios. But AMD would be wise to ramp up R&D spending a lot if Willamette proves formidable: Their upcoming K8/Sledgehammer must be at least as good in 32-bit x86 and x87 stuff.