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.
Politics : Formerly About Advanced Micro Devices
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Mani1 who wrote (106659)4/19/2000 1:56:00 AM
From: Petz  Read Replies (2) of 1574054
 
Mani, re:<As I said then there is no way Intel has a 1 million Coppermine/week run rate from only one fab.>

I think the sign was legit BUT...

This fab does 10K wafers per week 10K * 100 CuMines/wafer gives 1 million CuMines. BTW, 100 CuMines/wafer is less than 50% yield, not really very good. In fact, 100 CuMines/wafer is really more like 33% yield, which is REALLY BAD.

In its other 4-5 fabs running 0.18u process, Intel has the equivalent of 2 other of these 10K WPW fabs. So they should be making 3 million CuMines per week for a good week. (Note that so far as we know, that sign was only there on ONE good week.)

But they are NOT even making 2.5 million CuMines per week.
WHY?

Because they have a very severe packaging problem with the so-called "flip chip" or FCPGA package. The 2.5 million theoretical CuMines are cut to 1.5 million by the packaging problem.

Furthermore they are having VERY BAD BIN SPLITS. The 1.5 million that are successfully packaged are cut even more by the PISS POOR BIN SPLITS.

This would also be the explanation why CellarMines above 533 MHz are taking so long to appear. (Problems with FCPGA packaging and poor bin-splits.)

So, we start with 3 million theoretical CuMines per week.
Cut that to 2.5 million because MOST weeks aren't that good.
Cut that to 1.5 million becuase of the PACKAGING PROBLEM.
CUt htat to 1 million because of BAD BIN SPLITS.

So Elmer probably wasn't lying or misinterpreting the sign. The problem lies elsewhere. Furthermore, the fact that Intel can only occasionally get 1 million working CPU's out of 10,000 wafers is really proof that the yields SUCK big time, not that Elmer was lying or misinterpreting the sign.

Petz
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  

HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2025 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News