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: John Walliker who wrote (59114)	10/29/2000 2:47:52 PM
From: jim kelley	Read Replies (3) \| Respond to of 93625

I agree. There is no difference. I think the 820 was derated for product positioning reasons by product marketing at Intel. The use of an MTH was a bad idea from the beginning. If they had not derated the product and tried to go around the the RDRAM the 820 would have been a much bigger success IMO. This would have helped the RDRAM production ramp and brought prices down sooner. The recall of the 820 + MTH happened about the same time that Intel got noticed on DRAM and DDR memory and controller royalty issues( circa April). I wonder how this affected their decisions.

To: John Walliker who wrote (59114)	10/29/2000 2:52:20 PM
From: Bilow	Respond to of 93625

Hi John Walliker; Re power considerations and the number of banks open on RDRAM... If they had allowed the theoretical max number of banks to be active, they would have required active cooling (i.e. fan) in addition to the heat spreader. That's my understanding of the issue, but I don't see why it would have much to do with the i840 vs the i820. -- Carl

To: John Walliker who wrote (59114)	10/29/2000 2:55:08 PM
From: Dan3	Respond to of 93625

Re: In that case, I really don't understand what is going on. There was a lot of discussion of this issue last year. Here is a summary: The heat spreaders used on RIMMs can only dissipate the heat produced by a single RDRAM chip in active state. The result is the RDRAM granularity problem where you need a minimum number of RIMMs to keep enough chips active to support adequate latency. Early RDRAM specifications called for a "wind tunnel" which would have minimized this problem - but it turned out the "wind tunnel" wasn't feasible. Rambus wanted an enclosure fitted around the RIMMS and larger heat sinks together with a powerful fan to maintain a high airflow across the RDRAM heat sinks. As it turned out, the electrical requirements of the RDRAM buss was such that it dictated the location of the RIMMs on the board, and that location was not one that made building the "wind tunnel" and fitting adequately sized heat sinks possible. If the RDRAM chips and sockets are spaced widely enough apart to support finned heat sinks and fans, the motherboard traces become too long. The only solution for keeping enough RDRAM chips active to make performance competitive with PC133 (much less DDR or dual DDR) is to use multiple channel motherboards like the 840 - which are significantly more expensive and provide only a moderate performance increase. Dan

To: John Walliker who wrote (59114)	10/29/2000 5:58:17 PM
From: gnuman	Respond to of 93625

John Walliker, The only place I can think of where the power dissipation could depend on the number of active banks is in the memory chips/modules themselves. The same memory is used in the 820 as in the 840, so why the difference? I think you're correct. The thermal issues on the RIMM are addressed in two ways on the 82820 MCH. There is a thermal sensor on each RDRAM die that can trigger throttling back of Read/Write access to memory when preset conditions are met. Then there is the Pool A and B issue that defines the number of devices on a RIMM that can be in the "Active" power state. A lot of engineering went into thermal control of a RIMM. (Check out the 82820 MCH spec, pages 128 - 131). ftp://download.intel.com/design/chipsets/datashts/29063002.pdf Either situation can affect the performance of RDRAM. As for the i840, being dual channel with paired RIMM's there are twice as many devices in the A pool. And with the higher effective bandwidth, I suspect any performance degradation is less noticeable. JMHO's