Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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: Sarmad Y. Hermiz who wrote (219819)	12/9/2006 1:39:04 PM
From: dougSF30	Read Replies (1) \| Respond to of 275872

Furthermore, the "capacity gains from going to 65nm" seem to be limited, even in the ideal case, due to the die being nearly 70% the size of the 90nm equivalent. But with AMD defining "mature 65nm yields" as getting at least as many good die from a 300mm wafer as they do with 90nm (instead of the ideal of having an equivalent, mature fraction of good die per wafer), it is unclear that there is ANY capacity gain AT ALL right now, relative to 90nm on 300mm wafers.

To: Sarmad Y. Hermiz who wrote (219819)	12/9/2006 6:29:42 PM
From: economaniack	Read Replies (1) \| Respond to of 275872

Sarmad, if I were sensitive I might think you went looking for a post to embarrass me with. If that was the case perhaps you should work on your language skills. My post was in response to a suggestion that AMD might skip 65 nm altogether and move straight to 45nm design rules or was that beyond possibility. Hence "beyond beyond." I explained that 45nm was (for AMD) years away, while 65nm was shortly to enter production and was needed for capacity reasons "next year." Perhaps you don't understand how AMD ramps production in their fabs (it is different from Intel). AMD uses their production fab to develop the production process to acceptable yields and binsplits. Typically wafers in start at about 1000 a month (or less) at very low yields and ramp about 1000/month/month assuming yields and bins improve as expected. If they started a production ramp in June (a couple months after my post, as I then expected), those chips would finish processing in September, 3 months of the ramp should see something like 6000 wafer starts (total) complete to packaging by an early December launch which is about the minimum necessary for a limited (say low power desktop) release. Note that this rate has them at crossover (more 65nm wafer starts on 65nm than on 90 nm) in Jan or Feb (assuming the fab is currently running 15000 wafers a month).If they have achieved "mature yields", so that they are now yielding more usable chips on a 65nm wafer than a 90nm one then the output of the fab will grow pretty rapidly as yields improve to peak levels (another 40%-50%?) and they reach capacity in terms of wafer starts. The first couple times I thought that AMD would have huge capacity gains as they moved to new design rules, and was disappointed as the ramp always moved excruciatingly slowly. I now understand this as a feature not a bug of the AMD manufacturing process. They don't have the facilities to process wafers in sufficient capacity to refine to production yields in their test fab, and they cant afford to commit too many wafers to a new process before it begins to yield well, so their ramp looks much more like an exponential s that Intel's which produces meaningful volume earlier and continues over a longer time. Hope this helps e