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 : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: Alan Bell who wrote (28851)	9/7/1999 12:32:00 AM
From: Tony Viola	Respond to of 93625

Alan, My conclusion is that designing a Rambus board, particularly with a complete design guide, isn't all that hard. I agree with you, given that the mobo design group has high frequency board design experience. I wonder if Intel is doing any. They certainly are the best in my experience. Well, I think Intel might farm it out, but the end result is the best. Tony

To: Alan Bell who wrote (28851)	9/7/1999 12:47:00 AM
From: Bilow	Read Replies (1) \| Respond to of 93625

Hi Alan Bell; Regarding those Rambus PCB requirements. You note that right angle turns were disallowed in the Motorola ECL design guide many years ago. This is true, but it is also true that this has not been a recent design requirement for large number of signals on standard PC motherboards. ECL is known as a family that is hard to design with. In fact, having designed at a company that built supercomputers from ECL, I found it very easy to work with. On the otherhand, the guys who had to route the boards did complain a lot. So comparing rambus to ECL is not a good thing for rambus, from the point of view of board design. Do you agree that having traces who's width decreases at vias (instead of the industry standard of keeping them the same or widening them) is a new requirement for motherboards as well? Would you like to suggest that this is commonly supported by layout tools? Maybe name one? Regarding those holes in the ground plane. These have been disallowed for high speed logic for many years. What we have here is the addition of high speed logic rules to a domain that hasn't had to use them. I would like to note that the rambus parts have a very high pincount relative to their area. This is going to make puncturing the voltage planes a lot harder to avoid. In fact, guys I know having high pinout/sq. inch BGA packages on cards recently are finding them a lot more of a pain to route, and that is without all the controlled impedance and wire length requirements. Okay, controlling wire lengths is hardly a new technology, either. In fact, the Cray-1 was famous for it. But that was a highly expensive supercomputer, not a commodity box. On previous motherboard designs, the controlled impedance stuff would apply mostly to the clocks. The rest of the signals are left to autoroute. I don't think that these design requirements are all that bad, just that they do form added expense and trouble to the board maker. ECL had the same sort of troubles, that is why ECL was known as such a difficult family to work with years ago (and still is). As far as trouble for the board houses, we will have to see. It is clear that the trouble for the memory houses was already vastly underestimated, would you really be surprised if the box makers had the same sorts of troubles? The basic fact is that rambus runs a lot closer to the edge of failure than the alternate bus techniques (and hence the more complicated rules), but fails to deliver significant advantages. -- Carl