RDRAM, Intel, & Future

[scrumlord]

Look, I know RDRAM is expensive and proprietary, BUT it provides (to my admitedly sub-engineer level knowledge) higher performance and lower latency. Add to these the apparent fact that very high speed DDR/DDRII has frequency issues that affect stability. Why is everyone so anti-Rambus? Further, the specs on Yellowstone seem amazing - at levels of performance and efficiency not available with DDR based technology. If Intel abandons Rambus, is this a good thing? Is this anti-Rambus, pro-DDR hype a cost issue or is it a proprietary tech issue? I just want to build fast, rock-solid systems. Maybe my premises on Rambus tech being superior are wrong - anyone have opinions on this?

 

[wschuerm]

For the moment as in NOW rambus IS superior i don't thinkanyone can argue his way out of that fact. Yes rdaram is expensive... Yellowstone is a product on paper so that is not something i look at, what i have seen are tests on Nforce 2 and Granite bay, and these to take the wind out of Rambus's sails, GB performs equally good (4.2 GB bandwidth) and will be available this year, Nforce 2 (for AMD) is slightly better and gives a memory bandwidth of 5.2 GB using DDR 333. sis and via will also throw ual channel ddr 333 chipsets on the market in the coming months, so performance wise rambus's days are numbered because i do not know of any new ram tech they are marketing in the coming months.

But yes you are right for the moment RDRAM is superior, if you want to buy a top performing system today you'll have to go with RDRAM and the 3.06 PIV. But if you don't want to buy a system today then i'd wait, intel does not have any RDRAM based chipsets on its raodmap any more, springdale will replace it as high end solution for the Preston, so for the moment rambus days seem to be numbered, remenber being the fastest means nothing when you don't have support.

Remember the time You used 20Mb harddrives

 

[scrumlord]

Thanks for the reply. What about high-end DDR instability issues? I have been reading a lot of frequency issues, which in the past have been real deal breakers on pushing limits on technology. Mind you, this is not personal experience and is only hearsay from forums - I haven't tried DDR-400+.

 

[wschuerm]

ddr 400 can have issues but look at what the market offers there brands that support CAS 2-2-2 T2 timings and run stable there are PC 3500 stickson the market now (=ddr 433) i mean the stuff is OK, imo instabilty is caused by overclocking and asyncronous use of ddr 400, there simply isn't any chipset/cpu combination that can use this, sure theres via kt 400 and stuff but athlon only has 333FSB so some people buy ddr 400 and jack the fsb up from 166 to 200 i can imagie that this destabilises the machine, the only chip that can run syncronous with ddr 400 is the PIV A but there isn't any intel chipset supporting ddr 400 so there it is, only reason it exists is for overclocking, in the future, Q4 2003, the sprindale will support ddr 400 and an 800 mhz fsb, (for preston), here ddr will finally be usable in a stable platform without any OC's' it will be dual channel hence DUAL ddr 400 will provide the 800 the processor will be able to use......

Remember the time You used 20Mb harddrives

 

[scrumlord]

Thanks! Good explanation. I forgot to consider the AMD fsb issue.

 

[Crashman]

The thing about Dual Channel is it means you no longer NEED the fastest memory to get the best performance. Dual Channel PC2100 (DDR266) has 4200MB/s bandwidth, the same as RIMM4200, a pair of PC1066, and the P4 "533" bus. DDR400 becomes a non-issue with Dual Channel unless your overclocking your bus to "800" speed!

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>