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: Bilow who wrote (31484)	10/3/1999 11:44:00 AM
From: grok	Respond to of 93625

RE: <There was some talk about solving the signal integrity problem by requiring ECC, in addition to two RIMM slots. If this did become operative, and memories have had to use this technique in the past, it would add another 12% or so to the die penalty for RDRAM versus other technologies.> Carl, I hope that you are feeling better today. As for ECC, I don't think that it is a solution to Rambus' problems. It may be useful to discover failures more quickly but I don't think it can correct the problems and get them back on the air quickly. The standard ECC in use today uses 8 extra bits for a 64 bit data word and is capable of detecting any two failures and can correct any single failure in that 64 bit word. But if Rdrams are having trouble driving a particular trace then that trace gets used 4 times as the 64 bits are muxed over the 16 data pins (to be precise I should say as the 72 bits are muxed over the 18 data pins). So 4 failures would occur and, in general, ECC cannot correct 4 failures (in certain cases it can, check out Chip Kill). However, if someone gave me a Rambus machine (since I would not but one) it better have ECC because I would never trust it otherwise. But if it is failing the ECC won't fix it. It'll just tell you that it's failing rather than letting you find it by the blue screen of death.

To: Bilow who wrote (31484)	10/3/1999 12:23:00 PM
From: grok	Respond to of 93625

RE: <By the way, I've been getting the urge to write a Java applet that will allow you to simulate signals passed down a 3-RIMM Rambus channel, assuming that the MB traces and RIMM traces and termination have various impedances, which differ. The reason for this, is that Intel specifies (in a pdf file,) what kinds of tolerances the MBs have to be built to. I seem to recall that the MB impedance was 10%. Sure would be a cool simulation, but it sounds a little too much like work.> Carl, as you know interpretted Java is slow so I suggest that you get an Athlon machine to run your simulations. Also, be sure that it has Sdram so that you can trust the results.