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: Petz who wrote (86950)8/14/2002 3:51:28 PM
From: Pravin KamdarRead Replies (2) | Respond to of 275872
 
Anyone know what percentage of Intel's flash production is NAND? I know AMD does mostly NOR. Samsung could be putting some serious heat on anyone else making NAND flash:

eet.com

Pravin.



To: Petz who wrote (86950)8/14/2002 6:07:48 PM
From: ElmerRead Replies (1) | Respond to of 275872
 
If the rocket lots from the pilot production (say, 500 WPW) were good, they would begin reconfiguring the fab for 50% Hammer production. Until you get a good stepping you wouldn't do that. Four weeks dead time for half the fab sounds reasonable to me. Can Intel convert a fab from Willamette to Northwood in four weeks?

I really don't know what you're talking about. First off, I don't believe they are running 500WPW of non production Hammer unless the SOI process is in real trouble and still needs lots and lots of work. There just isn't any reason to run 500WPW of preproduction material when it only takes a couple of lots for systems validation and reliability qualification. Without the validation process completed, they don't know if the bugs are worked out and there's no way they're going to want to scrap all that material if it's no good. Additionally what if there's a reliability problem that needs a full layer fix? They can't just hold a bunch of wafers at metal like they would for a minor bug fix. If they're saying they'll start production in 4 weeks you have to ask yourself why are they waiting 4 weeks if the design is correct now? And if it's not correct now, how do they know for sure it will be in 4 weeks?

What is your experience?

My experience is that the design validation step is the most time consuming. A new stepping requires a full regression of the validation process and if they are saying that will take 4 weeks to complete then that means IF everything is perfect. It is possible that the 500WPW are production material that AMD is willing to risk scrapping if necessary, but if that were the case they'd have much more advanced demo systems out there than we've seen. Intel was demoing 4-way McKinley systems 1.5 years before the eventual release. What we've seen from AMD suggests they aren't close enough to production to be starting 500WPW risk material, imho.

EP



To: Petz who wrote (86950)8/15/2002 12:17:19 PM
From: burn2learnRespond to of 275872
 
it's possible they have a current stepping they know is good but need to do process tweaks for speed. targeting of implants, maybe a new process fix for salicide or poly change. In 4 weeks from starts you would get m1 etest data and be able to see if your process tweak worked and make a gamble that the gains would still be valid at end of line

you have two big buckets for speed, steppings (design) and transitor revisions (process)