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: combjelly who wrote (228928)	3/25/2007 10:28:54 PM
From: pgerassi	Read Replies (2) \| Respond to of 275872

Dear CJ; But that is the problem. Copy exact assumes that each tool model is interchangeable. That wherever scanner model 1234 was used in the development line, another scanner model 1234 would work as well. APM requires no such assumption of interchangeability. It works even if scanner model 1234 works differently for each one. In fact it can use the knowledge of which one is best for each step to route that step in all lines through that tool. If a new tool is better than the current top dog at a step, all production lines can be rerouted to use it at that step. Sarmad talks of wafer differences, but assumes like Intel that all the tools are the same. The variations in tools is likely to be higher simply because they have to be cutting edge. In fact as Intel normally gets the initial tool that comes off the production line, the will be even greater differences between it and ones that come off next. All you have to do is look at the first production CPUs off the line versus ones that come a few months later. The first ones generally use more power at lower clock and vary much more from die to die than dies that are produced later. Also that 2006 ITRS update shows that "double exposure" uses 8 steps compared to one with immersion. Given that 40 masks are used in the critical layers out of 200 to 300 overall processing steps. Using 320 steps with DE versus 40 steps with Immersion makes the 200-300 steps become 480-580 steps with DE. That halves the wafer throughput so it takes twice as many 45nm DE fabs to do the same output as 45nm immersion fabs. So AMD's 2 45nm immersion fabs would have the output of 4 Intel DE 45nm fabs. Along with Chartered, AMD might have as much 45nm capacity as Intel. There goes any problem with AMD not being able to supply as much as Intel. At 50% MS, there will be little, if any, ASP premiums for Intel. Let us see how Intel will do with its average ASPs of less than $100. Pete