SI
SI
discoversearch

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
RMBS 107.05+9.0%3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Alan Hume who wrote (18204)4/5/1999 12:04:00 PM
From: Dave B  Read Replies (2) of 93625
 
Alan,

Regarding the instability of DDRDRAM vs. DRDRAM, the absolutely 100% correct answer is that I have no idea why the president of Sony would say that DDRDRAM is unstable. I'm not sure I'd phrase the question as why one would be stable and the other not since it appears that Rambus DRAM is not stable yet either. Since neither is stable, I suspect it's a just question of just trying to get stuff to work at these high speeds.

Having said that, I have a few speculations based on anecdotal feedback from Herman (my ASIC designing friend). I may be botching these descriptions and comments completely, in which case the fault is mine, not Herman's. If anyone has any different views, please comment.

First, he once told me that the industry had a devil of a time getting 100Mhz SDRAM to work! With 100Mhz SDRAMs you have 10ns between key clock transitions and he said that the manufacturers were trying to get it to work with 7-8ns settling times for address and data lines. It just wasn't working and he says that Intel finally had to come in and drive the PC-100 spec, which spec'ed shorter settling times among other changes (it sounded as if he was saying that there are other differences as well between the PC-100 spec and other versions of SDRAM, but he didn't say that exactly) which then allowed the technology to work. I only bring this up because since DDRDRAM uses both edges of a 200Mhz clock, they're now talking about 2.5 nanoseconds between clock transitions, which is significantly tighter than the 10ns between transitions in 100Mhz technology. Assuming that they have to settle the address lines in 1.5 to 2ns, that's a significant reduction. I'm not surprised that they're having problems, just as Rambus is. In both technologies, we're moving down into picoseconds for rise and fall times for signals.

There's another difference in the technologies where Herman thought Rambus made and elegant improvement. However, I need to go to a meeting and reboot my system before I go, so I'll finish that thought in a couple of hours.

Good day for Rambus!!

Dave
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext