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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: pgerassi who wrote (106813)	4/20/2000 7:18:00 AM
From: Elmer	Read Replies (1) \| Respond to of 1573899

Re: "If I am off the mark, and I have yet to see where, (I gave Intel the benefit of the doubt), PROVE IT! Show me What the real NUMBERS are. From that analysis, from PUBLICLY released information, Intel is getting poor yields. Let's take it from the other direction." Pete you've got to be kidding. You know I'm not going to provide any data that isn't publicly available and as far as I can tell, Intel hasn't broken down the volume shippments for Q1 into specific products, so I can't help you. I will add some comments below: "1) From your message, Intel shipped at least 10 million Coppermines in Q1." I didn't actually say that but I definately implied it so I agree. "2) Yields are fine. I take this to mean, 1 shippable CPU for every 2 die, minimum" Sure but I'm not sure what you mean here, raw die on the wafer, sorted die, packaged die? Whichever it's true eitherway. Re: "3) Thus 10,000,000 * 2 die equals 20,000,000 die in Q1" I can't argue with your arithmatic but rather than 20 Million it could have been 10 million with 100% yield. In #2 you said 50% yield minimum. Your use of 20 million here is unwarranted. Re: "4) From public information, greater than 300 die fit on a wafer. (If there is less, tell me the correct number)" I'm not going to tell you the correct number, you'll have to buy one yourself, crack it open and measure it. And will you be sure that's the latest die size? What about a shrink? Re: "5) Thus, at most, 67K 0.18 micron wafers were used for CPU shipped Q1" If only 10 million were shipped and the finaly test yield worked out to only 150 DPW then yes ~67K is the number of wafers used to process the total number shipped in Q1. Re: "6) There are 14 weeks in Intel's Q1." OK Re: "7) Thus on an average 5K 0.18 micron wafers were processed[in Q1]" Yes and no. Not all units shipped in Q1 were processed in Q1 and not all units processed in Q1 were shipped in Q1. Some of those 10 million(or more) units shipped were processed in Q4. Some of the units processed in Q1 won't be shipped until Q2. Re: "8) To get 1,000,000 CPUs from one plant at week 7, that plant had to process between 3.3K and 6.7K wafers that week" Actually they had to sort that many wafers that week. Who knows exactly when they were processed, but no doubt not much earlier. Re: "9) Thus that plant should be able to supply 8 million good CPUs from week 7 to week 14. If it did not, this goes contrary to all statements of "We are supply constrained"." Two things here. First I never said that fab produced 1 million units each and every week. I only confirmed that it was capable of doing so, but for the moment let's assume you're correct. Those units would still not show up in the warehouse for about 6 weeks later and then not in systems for several weeks after that. That essentially brings you to the end of the quarter. Re: "10) From week 1 to week 6, it is at worst linear from zero to rate at week 7. Thus that plant should have produced 3 million more CPUs for a grand total of 11 Million" Not true at all. If the plant didn't come online at .18u until WW1 then there is still the TPT to finish those 1 million die (6-8 weeks). There may have been ZERO output prior to that week 7. Re: "11) I will make an assumption here: Intel shipped no more than 15 Million Coppermines in Q1" No comment here. Re: "12) All other plants produced, at most, 4 Million Coppermines for Q1" No comment here. Re: "13) Given 50% yield, they processed in Q1 27K wafers" Wrong again. You are forgetting fab TPT. The units shipped in Q1 were not all the result of units processed in Q1. Only some were. Some were from Q4'99 and many from Q1'00 haven't seen the light of day yet, especially late Q1. Re: "14) At 50% conversion end of Q1 per CC and assuming 0% conversion at beginning of Q1, (believe this to be worse case), they can only produce 4K 0.18 wafers per week at end of Q1" You're really starting to fall apart here because of your errors in assumptions. Re: "15) Thus capacity of Intel end of Q1 is 10K 0.18 wafers per week" All based on faulty assumptions. Re: "16) Thus total capacity of Intel is at most 20K wafers, any kind, per week" Waaaaaaay off the mark. Re: "17) Intel has about 3.3 times the current capacity of AMD Fab 25. NO WAY! 18) Intel claims to be able to process 60K wafers per week (10x AMD by all accounts). 19) Intel should be at 30K 0.18 wafers per week now." Now revisit your calculations with the knowledge that: #1 current wafer starts won't be seen at retail for ~12 weeks. #2 Assume much higher yields. #3 Intel started shipping GHz units about 1 month ago. #4 What improvements might Intel have put in place since? #5 When did Intel say the current shortages would be over? EP