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 : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: richard surckla who wrote (71845)	2/18/2002 1:37:33 PM
From: Petz	Read Replies (1) \| Respond to of 275872

I consider Tom's "conversion" to favor RDRAM to be a temporary phenomenon caused by excess demand for DDR driving the price close to that of RDRAM. Hopefully, this will cause Intel to re-convert to the RDRAM religion again. That would be the best for AMD. The best DDR boards for the P4 are within 4% of RDRAM boards on memory intensive applications and give better-than-RDRAM performance on latency-sensitive applications like Office. DDR is still cheaper, so why go for RDRAM? On the athlon, all we need is 167*2 MHz FSB CPU's to get additional performance out of already-available PC2700 DDR. Dual channel DDR boards will be available for both P4 and Athlon long before RDRAM doubles it's frequency, but higher frequency RDRAM will cost a LOT more than PC2100 or PC2700 and won't even have a theoretical bandwidth advantage once DDR dual-channel is available. The only possible advantage I can see to RDRAM is if there is if it's latency when using an on-CPU memory controller could be brought as low as DDR with an on-chip memory controller. Ali Chen can probably enlighten us as to the probability of that occuring. Petz

To: richard surckla who wrote (71845)	2/18/2002 5:34:04 PM
From: eCo	Read Replies (1) \| Respond to of 275872

Ed Stroligo on "Tom's Conversion": "Wham, Bam, Thanks, RDRAM? Ed Stroligo - 2/17/02 There Is No Getting Around Rambus???? I don't know why Tom Pabst lets these guys write articles for his site. The article essentially says that RDRAM is likely to make a comeback since DDR can't quite keep up. At the moment, for most things, this is dubious. To the extent that it's true (and it's only makes a sizable difference in the case of a relatively few bandwidth-intensive applications/games), the real advantage will emerge only when Rambus introduces PC1066 and Intel (or overclockers) approach 3GHz. Now if there were nothing on the DDR side to counter that, the authors might have a good point in a few months. Unfortunately for the authors, there will be something to counter that: dual-channel DDR. Via is supposed to come out with the P4X600A, which is supposed to include it, as early as next quarter. Intel is certainly thinking along the same lines with the Granite Bay chipset to be introduced a quarter or so later. This is part of the reason I'm not getting all wound up about PIVs quite yet. If you have to make your purchase count for a while, you probably want to wait until that milestone is reached. On paper, dual-channel DDR266 provides exactly the same theoretical bandwidth as dual-channel PC1066, so unless implementation is poor, looks like a Column A or Column B choice to me. If DDR333 gets worked out, the paper advantage shifts to dual-channel DDR. Of course, we haven't seen any of this, but then again, we haven't seen PC1066 either. Poor Work It is never a good idea to express an opinion and then put in the facts that fit and leave out the ones that don't. That is what this article looks like. There are some good reasons to be at least somewhat skeptical about dual-channel DDR. Via and Intel could be delayed. There could be technical difficulties, especially with getting two sticks of DDR333 to play together. But there are no good reasons not to mention dual-channel DDR at all. All that does is make the informed reader ask "Why not?" Ignorance is not bliss here, and those inclined to think the worst will. overclockers.com Hehe...THG's article simply means they're through shorting RAMBUS, and are now going long (despite explicit claims to the contrary). eCo