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: wbmw who wrote (237248)	7/26/2007 1:22:55 PM
From: dougSF30	Read Replies (1) \| Respond to of 275872

EDIT: Wow, or 28% yields with .5 def/cm^2 itself. Don't get suckered by 159. "wafer yield" can be expressed in NUMBER of good parts, not % of good parts, if you're AMD :) ------------------- Per Elmer, Intel's yield on Penryn QCs (107mm^2) is 77% (3 times as good!) for 450 net die, or: 225 MCM QC Penryns / wafer AMD gets 58 Barcs. :)

To: wbmw who wrote (237248)	7/26/2007 2:40:53 PM
From: Reseller Mike	Read Replies (2) \| Respond to of 275872

Yields... I find it utterly amazing that people that can't believe a word from AMD about any of their presentations or charts or schedules suddenly believe that the yield number on one of those said 'coils' is trusted. It does say less than 0.5 so for we know it could be .35, .30 .45, .25 or whatever. Let's face it, there is only circumstantial 'evidence' of the yield and binning from AMD (or intel for that matter, despite what Elmer posts). I think that the demo today of the 3.0G Phenom is a good data point that indicates the process can hit frequencies that will be needed. What it took to get there we don't know and what's left to fix is still to be determined. Regards.

To: wbmw who wrote (237248)	7/26/2007 4:05:46 PM
From: jspeed	Respond to of 275872

wbmw - thx for the correction. 30%-ish is more in line with my expectation. Turns out I converted defects/cm^2 to defects/in^2 incorrectly.

To: wbmw who wrote (237248)	7/27/2007 6:19:01 AM
From: Rink	Read Replies (2) \| Respond to of 275872

Wbmw, re: Yield of 32%. Some parts of the K10 die are designed with redundancy in mind to increase yields, like probably the larger cache blocks. This means yield is probably higher than you calculate (using your own presumptions of course). I'm not counting on it but maybe it'll still be possible for AMD to salvage part of the defective QC's and sell them next year as DC's. Regards, Rink

To: wbmw who wrote (237248)	7/27/2007 3:29:49 PM
From: Petz	Read Replies (2) \| Respond to of 275872

re: Barcelona is 283mm^2, so 16.823 on a side. Use 300mm wafers, and 0.45 defect/cm2: AMD's yields are 32%. Better than that. Defects in the cache areas don't affect yield due to redundancy. L2+L3 takes 1/3 of the die area xtremesystems.org so the die area susceptible to single defect failure is 189mm. Using .0045 defect density, we get a yield of 0.453581617 Hmmm, and I predicted 46%! LOL, and here's what nuclear said, It should be obvious that 40%+ YIELDs for a large die is pure fantasy concocted by someone who lives in fantasy island. Petz