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: misen who wrote (219190)	12/5/2006 11:42:21 PM
From: combjelly	Read Replies (2) \| Respond to of 275869

"This is the basis for the simple Poisson model." Which is the classical model. But there is another dimension to any good model below the 130nm node, and probably the 180nm one. Bin splits. Now traditionally bin splits were a separate category, but it shouldn't be any more. Because what good is a die if it can't function at a reasonable clock? Sadly, there are a lot more factors than just particle distribution. How closely the image approximates a rectangle is only one. Especially as manufacturers move beyond a Manhattan layout.

To: misen who wrote (219190)	12/5/2006 11:52:56 PM
From: Sarmad Y. Hermiz	Read Replies (1) \| Respond to of 275869

misen, you sound like you know what you're talking about. So I hope you won't mind sharing some knowledge. Do different transistor designs yield differently just because the designs are different ? Another way to ask this question is: Are some designs more susceptible to a fatal failure just because of their design. i.e. their lines are so thin that an atom being in the wrong state causes failure, whereas another design has the lines thicker, so a one-atom defect doesn't make it fail ? Or, is it that all 65nm designs have the same susceptibility to defect density ? (for instance, if their lines were thicker, they wouldn't be 65 nm). And, do differences in fabrication process steps (between Intel and AMD) lead to different defect density ? ALso, do different designs bin differently ? such that if the lowest acceptable clock speed is say 2.0 GHz, some parts will operate correctly at no higher than 1.8 GHz, and therefor are not "defective", but still, are not good enough to sell ? Thanks, Sarmad

To: misen who wrote (219190)	12/5/2006 11:57:24 PM
From: Elmer Phud	Read Replies (2) \| Respond to of 275869

Misen, Nice to see someone talks my language:) You may argue with the assumptions above, but the assumptions are close enough to reality that the example represents real behavior. Yes it does. Let us remember that these are models and their usefulness is in predicting yield. Based on the assumptions you've listed, if you know the defect density for a process based on a current product or other process monitor, you can use the poisson or murphy's model to predict the yield on another new product for which you have no data. If your process is defect limited as you stated, then it does an excellent job of predicting yield on a die of any size. That's why we can say that if we know the defect density and Company X has a fab capacity of Y wafers and a die size of Z, then they should be able to produce n Good die per unit period of time. Question for you. What do you consider world class defect density? I have been saying it's .25/cm2. What is your opinion?