Re: Does AMD have yield problems? You tell me.
Could be. And you didn't even take into account that AMD also has two additional small, older FABs. Now lets use your system and analyze Intel.
Intel has what, 18 FABs? 2 at 10,000 WPW, 2 at 8,000 WPW about 8 at 6,000 WPW and 6 at 2,000 WPW. It's something like that, isn't it?
So in 13 weeks there were 13*96,000 = 1,248,000 wafers manufactured - is that about right? And Intel got something less than 40 million PIIIs, Xeons, and Celerons out of them. Which works out to about 32 good die per wafer. Since the new PIII stepping is on 92mm2, while AMD has to crank out 120mm2 chips, yields must really be bad at Intel, eh?
It may be irrelevant, but I have this notion that Intel makes more than just X86 CPUs at its FABs, so maybe yields aren't quite that bad. But if we take that into account, then maybe you should modify your "system" to take into account AMD fabbing logic for Vantis under contract, fabbing analog com chips for legerity under contract, fabbing embedded CPUs that it sells, and fabbing networking chips that it sells. All this in addition to the Classic Athlons, Thunderbirds, K6s, and Durons it makes. AMD continues to be the sole source contract FAB for two other companies, in addition to making many products of its own. Kind of like TSMC or UMC.
Just a thought, maybe those other things don't make a material difference and Intel is at 32 good 92+mm2 (the plus to take into account the relative handfull of larger Xeon chips) die per 8 inch wafer, while AMD is at 75 120- mm2 good die per 8 inch wafer (the minus to take into account the smaller Durons and handfull of K6's).
:-)
Dan |
