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 95.57+0.7%Nov 28 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Tenchusatsu who wrote (33481)10/31/1999 10:46:00 PM
From: Dan3  Read Replies (3) of 93625
 
Re: Show me the performance numbers that prove DDR is going to be such a phenomenal leap over PC133.

DDR is little different from PC133. Because of this similarity, it is straightforward to predict performance changes. Its performance can be expected to be identical to PC133 in all aspects except that it will transfer data at twice the rate of PC133. Its latency will be the same, its behavior during various patterns of same page/different page and open page/closed page access should be the same. But it will cut in half the time needed to transfer a line of data to the cache. It can be expected to cut 15ns from a 32 byte cache line access and 30ns from a 64 byte (Athlon or Itanium) cache line access relative to PC133.

Compare that to the best case PC800 difference of -2.5 to +5ns (depending on CAS 2 or 3) relative to PC133 on a 32 byte cache line (rambus is better than PC133 for a 64 byte cache line CPU, but falls further behind DDR on such a machine).

And that takes into account neither the addition of Virtual Channel technology, which is planned, the additional 1/2 clock savings that is available during some accesses due to DDR, nor the fact that PC800/40 is the top binsplit of the 9 grades of Rambus - the other 8 binsplits of rambus are all slower. Every one of these factors increases the advantage of DDR over rambus beyond what I have described.

And DDR doesn't need 25 to 35% more die space and requires only 1/3 the clock speed (making it much cheaper and easier to manufacture). Given that more memory is often more important to performance than faster memory, the fact that the faster memory is also considerably less expensive, and so allows the same budget to procure machines with more memory, is equally important.

Sorry if I'm getting a bit wound up on this, but I really think that an attempt is being made to quash a good, inexpensive technology and substitute an inferior, more expensive technology - and I find that appalling.

I don't care if some marketing types at Intel have made a bad "bet" with a billion or two of the company's money on Rambus and are desperate to get a return on it. As a computer consumer, I feel that's their problem, not mine. It's chump change to Intel, anyway - think of the taxes the writeoff will save.

Regards,

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