SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?


To: niceguy767 who wrote (89377)1/25/2000 1:26:00 PM
From: Dan3  Read Replies (5) | Respond to of 1571040
 
Re: the 800's have yet to materialize in any meaningful number in any meaningful place...

There are, I believe, considerable numbers of higher speed Coppermines being sold by Dell, HP, IBM, etc. The appearance (or lack thereof) in pricewatch is due to insufficient supply, not no supply.

What has happened with supply makes perfect sense if you think about what has been said by Intel.

Intel began ramping the conversion of its CPU FABs from .25 to .18 late last year, and expects the conversion to be complete by the end of this year. It's supposed to be a steady rate of conversion, so let's call it 20% per quarter for 5 quarters (at 5 FABs). AMD has two FABs, and both are now fully converted to (or started at) .18 or better.

I think that Intel's FABs can handle more wafer starts than AMDs (8K/week vs. 6K/week) but let's treat that as noise.

Right now AMD, with 17% of the market, and far less than that at the high end, has two .18 FABs (I know, Dresden isn't ready to run more than about 30% right now, but since Austin can produce more than are needed right now, so what?) while Intel with the rest of the market, and traditionally all but a few percent of the high end market has the equivalent of 1 .18 FAB.

Intel had planned to have the sweet spot at around 500MHZ for most of this year, with 600 to 700 being "high end" gradually rising during the year (remember their earlier roadmaps)

How the hell a near bankrupt AMD ever managed to turn itself into a larger company than Intel (at least as far as .18 FAB capacity is concerned) is beyond me. True, Intel spent $4 Billion in the second half of last year buying and digesting other companies, and spent money, time and energy trying to force a reluctant industry to switch to Rambus, so, clearly, upper management couldn't be bothered with supporting the CPU guys, but it's still quite amazing.

So, PB, Elmer, and Tenchusatsu, with little evident support or attention from upper management still managed to develop both competitive designs and the techniques to manufacture them, but Intel upper management would spend only for a glacially slow FAB conversion. (how pissed off must those guys be?)

So right now, AMD has a 6 to 9 month or so lead in production capability. I call it 6 to 9 months, because I think that by then Intel should have enough of its capacity at .18 that it will be back to downbinning (like AMD does now) to satisfy the low end - that last 20% to 40% is vital since it reduces production costs, but isn't needed for speed.

Thing is, 9 months from now, AMD should be ramping up its .13 at Dresden - and they'll have been picking up high end market share all year long.

And it didn't hurt that AMD's designs are good for 700MHZ at .25 Aluminum, vs. Intel being at 650 at .25 Aluminum, and, of course, AMD will be at .18 copper when Intel is still running significant wafer starts at .25 Aluminum.

AMD's main problem right now is that no one really believes that the current situation is possible - a genius team could conceivably have designed a somewhat better processor, but that the manufacturing capabilities of Intel, so famously unparalleled for the last 20 years were just blown past by "always screws up" AMD!

So it is taking time for the OEMs to add SKUs at the high end for AMD, and it will take more time for endusers to start buying them in volume, but AMD clearly has given itself the time it needs. It's been offering about 75MHZ more at any given price than Intel, because it can do so for the foreseeable future, and that's been feeding AMD's steady march up to the high end.

Regards, (and sorry for the long, opinionated post)

Dan



To: niceguy767 who wrote (89377)1/25/2000 1:36:00 PM
From: Paul Engel  Read Replies (2) | Respond to of 1571040
 
SPryGuy - Re: "Just look at the ease with which AMD's Athlon has gotten to 800MHz "

Yep - all it took was for AMD to Jack Up the Voltage by 0.1 volts !

Guess how AMD is going to get to 1 GHz - a "0.2 volt Juicer - maybe 0.3 volts !"

Paul