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.
Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?


To: Elmer who wrote (84686)7/12/2002 12:30:20 AM
From: muzosiRead Replies (2) | Respond to of 275872
 
Note to thread. Pete Gerassi claims Intel has 200,000 wafers per week capacity
i couldn't find it myself. could you post a link to that claim? just curious to see how he came up with that number.



To: Elmer who wrote (84686)7/12/2002 12:46:01 AM
From: wanna_bmwRespond to of 275872
 
Elmer, Re: "In 2001, TSMC claimed a 5.2 pct share of the world's semiconductor capacity, compared with the 6.1 pct posted by Hynix Semiconductor and 5.1 pct of Intel Corp, he said. Ross said TSMC is expected to raise its total monthly capacity to 381,000 8-inch-equivalent wafers at the end of the year from 332,000 at the end of March."

Looks like pgerassi was only off by a factor of 2.4x. That's almost how far off he was in his Thoroughbred scalability estimates from this post.

"Since 0.25u to 0.18u allowed for 73% to 80% speed increase and Intel has at least 40% gain from 0.18u to 0.13u, AMD should get at least 20 to 30% from 0.18u to bulk 0.13u (1.8GHz up to between 2.13GHZ (2600+) to 2.33GHz (2900+)). IMHO, if they would stay with bulk, they would get ultimately about 50 to 60% (2.67GHz (3500+) to 2.8GHz (3700+)). If SOI gets 20% above these, 0.13u SOI would top out at 3.2GHz to 3.33GHz (4300+ to 4500+)."

Message 17279208

Maybe that should be a new law for pgerassi. He can be a pretty good indicator, and all you have to do is divide his projections by 2 or more. It could be called Pete's Law. <ggg>

wbmw



To: Elmer who wrote (84686)7/12/2002 2:03:17 AM
From: PetzRead Replies (1) | Respond to of 275872
 
Well, well. Let's do the math assuming TSMC's estimates are right. First of all, TSMC probably underestimated Intel's "8 inch equivalent capacity" since they want to claim to be#1 in 1-2 years.

According to TSMC, they had 332,000 WPM in March, rising to 381,000 at end of year and Intel has approximately the same capacity. That gives and average of 350,000 wafers per month x 12 months = 4.2 million wafers total. Intel capacity should be rising even faster than TSMC's, because 12" wafers have 2.25 x the surface area and suffer lower edge losses than 8" wafers. 12" wafers are a low % of TSMC's capacity but are supposed to be over 50% of Intel's by year end.

But I'll stick with the 4.2M equivalent 8" wafers number anyway.

Since 75% of Intel revenues come from PC and server CPUs, let's assume that Intel uses 60% of its capacity for PC CPUs. You'll have a hard time convincing me that they waste more than 10% of capacity each on chipsets, flash, comm/networking and embedded. That's 2.5 million wafers to produce no more than 150 million CPUs, or only 60 CPUs per wafer.

YUK!

That's WORSE THAN AMD! What was it, Dresden with 4200x13 wafers made >4M CPUs. Thats >73 CPUs per wafer. Austin had only 50-60% of that capacity and made almost as many Durons.

I guess it really only is AMD that has "world class yields" AND the "world's highest yields."

Jerry was right.

Petz