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 88.13+1.0%Nov 21 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Tenchusatsu who wrote (29709)9/16/1999 10:57:00 PM
From: grok  Read Replies (1) of 93625
 
RE: <In short, servers have huge amounts of memory. This requires interleaving, and interleaving has additional benefits. (For example, 460GX, the chipset that will support 4-way Merced, uses SDRAMs which are highly interleaved.) Because of this interleaving, page hits are considered to be very rare in a server, and thus the paging policy is to close (precharge) pages right after they are accessed, as opposed to keeping them open. This makes Virtual Channels useless.>

Doesn't the logic go like this:
1. Servers need high bandwidth
2. Interleaving gives high bandwidth
3. Interleaving requires constant access time from all the chips being interleaved so page hits on some of them cannot be used.

but it is not correct to say that interleaving is used because of huge memory. Maybe buffering is needed but not interleaving. In fact, won't DDR or DDR-II make interleaving unnecessary since they provide all the bandwidth that the chip set can deal with? And in a non-interleaved system, page hits can be used. So VC DDR would be great for servers unless Unclewest calls in an air strike.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext