To: John Walliker who wrote (40029 ) 4/15/2000 6:37:00 PM From: Bilow Respond to of 93625
Hi John Walliker; Re tight process &c. The stuff you are referring to is stuff that had to be included in order to improve manufacturability. While it undoubtedly increases yield a lot, it doesn't get the yield up to as high a result as a design that doesn't need it. If SDRAM needed this sort of stuff they would have included it. They didn't need it, so there is no reason to put it in. But RDRAM has to have it. The question is it enough to get the yield to SDRAM levels? If it were, you would think that SDRAM would be designed with that sort of thing, as such a technique could then clearly be used to slightly increase the already high yields for SDRAM. But SDRAM doesn't need it. When people break their legs, they get crutches. Crutches help them move around. Maybe they can move around almost as well as people without broken legs. But if crutches were so great, why aren't we all using them? The calibration stuff is a crutch to improve yields. It is there for a reason, and that reason is to help increase yields (as we all agree). There is no way that a crutch is going to make that process yield better than a design that doesn't need a crutch. If it did, everybody would put that crutch in all designs, which would just raise the yield bar a little higher. SDRAM doesn't have to calibrate its output strength. Besides, there are a lot more FET and process characteristics than output current. The yield problem is a complex, multi-dimensional problem, and it doesn't have a simple solution to achieve the yields that looser designs achieve. If it did have a solution, it would have other costs, such as performance, die area, etc. There just aren't any free lunches in engineering. -- Carl