wbmw,
but why do you suppose that 40% of their #1 CPU producing fab wasn't producing CPUs?
I am on a thin ice here since I don't really know how a fab operates, so I can only stay on a very general level.
My theory is that AMD bought equipment as late as they could get away with, and the last of the batches of the equipment were qualified for .13u production.
I don't know how easy it is to flip a switch, bring different masks and go from processing .18u to .13u and back. My assumption is that it is not trivial. Suppose AMD figured out that they could use additional 1000 WPW worth of production in 4th quarter. Would it make sense for AMD to qualify a newly installed line for .18u, then to run say 2 months worth of production, and the to start qualifying the line for .13u. Or would it make more sense to use the time in Q3-Q4 timeframe to qualify this new line for .13u production <edit> and forget the additional 1000 </edit> ? I think that's what happened.
Was it because they were manufacturing .13u Thoroughbreds that weren't set to launch for another 2 quarters?
This is what Hector said: The conversion to 130nm in FAB30 began in the fourth quarter, and the latest silicon output on this technology has already yielded more die per wafer at lower cost than the record numbers that we just achieved on 180nm. We have already(?) sampled to customers and expect to ship 130nm product for revenue this quarter. We are on plan to achieve full conversion by year end. In addition, we started silicon in the fourth quarter on our leading edge 130nm SOI technology. This is a technology that is -- will be used for our first lots of 64-bit Hammer products, where first silicon out is imminent.
So if AMD started processing .13u wafers in Q4 (we don't know when, so let's assume mid Q4), there has not been almost any packaged output prior to February. And the volume is only very slowly increasing as we speak, so that it the product can be launched early Q2.
That wouldn't make sense, unless you think they were stock piling, in which case you must also be from the camp that believes that AMD is still holding back processor releases because they feel comfortable with their current competitive position (not to mention their current money losing position).
I don't know where there is any room for stockpiling based on Hector's statement. As far as faster speed grades on .18u, I think AMD is squeezing the last performance increases of their .18u process, and if they deliver on their roadmap of XP-2200 which would imply 1.8 GHz, I would find it impressive.
Otherwise, the only other logical explanation to your theory is that AMD was demand limited in Q4,
AMD was definitely demand limited for Duron, very likely demand limited for Thunderbird, and possibly supply limited for XP. Remember that XP was introduced in October (=Q4), and if 1/2 ofAthlons sold were XP (> 2 million units), it would be an excellent ramp of a new processor.
Joe |
