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: StockMan who wrote (31354)	4/7/1998 11:34:00 PM
From: D.J.Smyth	Read Replies (2) \| Respond to of 1573075

his method at arriving at the yield needs work since (a) the unknown is the time it took to produce 10% of the chips using the .25 process as this information was not given and (b) if .25 produces 1.4X more the yield will eventually increase over the .35 in time

To: StockMan who wrote (31354)	4/8/1998 12:10:00 AM
From: Maxwell	Read Replies (2) \| Respond to of 1573075

Stockman: I see how your brain think now. Let me give you a problem. """""""Calculate the surface area of a cow."""""""""" Stockman's answer: Assume a cow is a SPHERE of radius R, then the surface area of a cow is S=4PiR^2 Maxwell

To: StockMan who wrote (31354)	4/8/1998 2:58:00 AM
From: Petz	Read Replies (3) \| Respond to of 1573075

S, Exciton's 8% yield estimate on Intel forum is faulty, here's my own plus an estimate of 2nd quarter 0.25 micron production: 1. Exciton, you forget that it takes 2-3 months, lets say, 10 weeks to fabricate a CPU in the complex 5-layer AMD process after the "wafer start." 2. From mid-October to mid-January, wafer starts were mostly at the SDC facility, not Fab25, started at close to zero wafers per week and ended at no more than 500 wafers per week. (Max capacity used at SDC was 500 wafers per week). Probably another 1K wafers were started at Fab 25 in January that were completed in the first quarter. 3. If we say they averaged 250 wafers per week, then over 13 weeks they started 3,250 wafers plus the 1,000 started at Fab 25, for a total of 4,250 wafers, and average of 360 wafers per week. There's 350 candidate CPU's per wafer, so thats 1.49M potential CPU's. 4. "More than 10%" of a little over 1.5M CPU's means there was 160,000 0.25 CPU's sold. But I'm pretty sure they also said that there were "hundreds of thousands" of K6 0.25's produced, so taking 0.2M as the right number, the average yield is 0.20/1.49 = 13% 5. 13% sounds pretty stinky, but what matters is what the CURRENT yield is. At least 3 analysts have stated in their reports that it is close to 50%. This is confirmed by the statement in CC that greater than two thirds of the 0.25 CPU's were from Fab 25. (0.67200,000 / (1000 wafers 350))= 38%. 6. If we assume the analysts are right, then for Q2 there will be 4 times as many K6's from 0.25u process because of the yield improvement (50% vs 13%), and 4 times as many again because the 'wafer rate' is now 1500 per week compared to 360 per week. (Actually its the average wafer rate from mid January to mid April, and thats probably only 1000 per week, a 2.8:1 improvement.) 7. All told then, we should expect about 0.20M x (50/13) x (1000/360) K6's produced using the 0.25 process in 2Q'98, or roughly 2.1 million. Petz