Will 'RIMM 4200' be 32 bit or 16 bit datapath?

[fkloster]

This will be the memory that fits in Asus' P4T-E 533...

 

[AndyHui]

That's a 32-bit datapath RIMM.

 

[dexvx]

Will it be backwards compatible with PC800?

I really hate it when things arent backwards compatible (wow to AMD for their backward compatibleness of Socket A !).

 

[AndyHui]

The P4T-E 533 and the P4T1066 will accept both 16-bit RIMMs and 32-bit RIMMs.

You can install the 32-bit RIMMs singly, but the 16-bit RIMMs still need to be in pairs.

 

[ViRGE]

Doesn't this seem a bit backwards? RIMMs were touted for their serial nature, which was/is supposed to be superior to the DIMM's parallel nature. However, with 32bit, we're already up to half the width of a DIMM. Mabey this is a "best of both worlds" thing, but I'm starting to wonder if mabey parallel really is the way to go with RAM.

 

[AndyHui]

We are up to half the width of SDRAM, but look at how much bandwidth you get over it.

 

[MadRat]

Plus RDRAM keeps the house so much warmer in the winter time!

j/k

I'm liking RDRAM technologies ability to ramp performance far faster than DDR technology. Maybe DDR2 will bring back the excitement of DDR technology, but as it is now the DDR is going the way of the dodo.

 

[ST4RCUTTER]

I think you can count on DDR being the industry standard for at least the rest of 02' and the first few months of 03'. This will be especially true when Intel introduces a dual channel DDR chipset. As it stands with current DDR speeds, PC2700 in dual channel mode will offer more bandwidth than the P4 can use and enough to keep RDRAM in it's niche-like place. The fact that the largest OEM (Dell) is no longer using RDRAM on the desktop doesn't help. DDR SDRAM is currently at a price premium, but you can bet your mass that it won't stay that high. High prices reduce demand, and eventually that will add up to inventory in the channel...you see where I'm going with this. By the time RDRAM can achive high enough speeds it will have to compete against DDR-II.

 

[SocrPlyr]

St4rcutter
where did youg get the idea that dell isn't using rdram anymore, i just checked and the Dimension 8200 Series their "cutting edge performance" pc still uses it... weird...
anyways i am finally happy to see the new 32bit rimms... basically all it does is put the dual channel rdram into a single board, much less complicated... it is a shame they didn't start this earlier... performance wise it is the same but just less of a pain...
contrary to all the crap that people are saying think it will be a long long time before we see rdram go away...
the mention of the dual channel ddr from intel is a nice thing to have but you are back to pair... a pain...
oh well these are just my thoughts
Josh

 

[dexvx]

ST4RCUTTER:

I dont know what you're thinking, but to me its amazing how 2 year old technology (DDR RDram) can keep up and beat brand new technologies like PC2700. I think RDram isnt good for servers (where amount means more than performance).

As for desktop Dell giving up RDram, their top of the line 8200 series all use RDram.

 

[andreasl]

To answer the original question. The new RIMM4200 modules will contain 2 channels of 16-bit PC1066 RDRAM chips, making it essentially 32-bit. The module itself is 232 pins vs previous 184 pins (I think) for PC800.



<< Doesn't this seem a bit backwards? RIMMs were touted for their serial nature, which was/is supposed to be superior to the DIMM's parallel nature. However, with 32bit, we're already up to half the width of a DIMM. Mabey this is a "best of both worlds" thing, but I'm starting to wonder if mabey parallel really is the way to go with RAM. >>



RDRAM is serial in nature but it isn't serial per se. Does that make sense? It uses a narrow channel which allows for much higher clockspeeds in the chip to chip signaling. The chips themselves are very parallell internally and the DRAM cells are the same as in SDRAM, DDR etc. This is how they manage to get so much bandwidth out of it. And it allows them to scale the bandwidth up very fast by both increasing the clockspeed and adding more channels to the module. In a few years there will be modules with 4 channels on them and even higher clockspeeds. This assuming they haven't been driven out of the market by then which would be a shame.

Of course there is more to RDRAM than that, but it's a start

 

[imgod2u]

I'm kinda confused now. What would be the difference between the Asus P4T-E 533 and the Asus P4T1066? Also, when would we expect 32-bit RDRAM modules to be out by? Will there be a dual 32-bit RIMM solution? When is the P4T-E 533 or the P4T1066 suppose to be out?

 

[ElFenix]

<< This assuming they haven't been driven out of the market by then which would be a shame. >>

not it its by an open serial memory technology.

 

[andreasl]

<< << This assuming they haven't been driven out of the market by then which would be a shame. >>

not it its by an open serial memory technology. >>



What I meant was that I want RDRAM to coexist on the market with an open standard solution like DDR and DDR-II. This way the consumer (me!) gets to choose which memory to use. For that to be possible there has to be at least 2 choices... Picking SDRAM over DDR is not an option IMHO.

But please, don't turn this into another DDR vs RDRAM flamewar....

 

[Dug]

Yes let us know about the Asus P4T1066. I've read the press release about the P4T-E 533, but have never heard anything about the P4T1066, until now.