Pentium 4 + PC1066 RDRAM at Ace's Hardware

dowxp

Diamond Member
Dec 25, 2000
4,568
0
76
doesnt look like rdram is going anywhere. 32bit looks interesting as it seems like an easy transition
 

Wingznut

Elite Member
Dec 28, 1999
16,968
2
0
Nice find, NFS4!


<< ...Rambus technology itself is quite promising. With the information that we have right now, it is probably the only memory technology roadmap that realistically will be able to keep with AMD's and Intel earth scorching pace of increasing clockspeed and CPU performance. >>

I believe that's the first time I've seen anyone say that.
 

fkloster

Diamond Member
Dec 16, 1999
4,171
0
0
Nice find NFS4. Rambus memory dead? Doubt it, Rambus inc. may go tits up but with PC-1200 (i.e. 64 bit / 600mhz) rimms on the distant horizon, all I can do is dream about marrying up that crap with a 2.8 ghz Northwood and drool helplessly... :p



<< it's likely that Intel's 0.13um Pentium 4, Northwood, will have a 533MHz FSB >>




Oh man is October gonna be a sweet month :)





<< You might be surprised by the significant reduction in latency seen with PC1066 over PC800. As addresses are sent 33% faster and the data returns from memory 33% sooner, latency is, according to Cachemem, 30% lower >>



Holy cow!! Latency on 533mhz rambus lower than SDRAM!!!!!!!



<< At the end of 2003, Samsung will introduce 64 bit RIMMs, which are essentially quad 16 bit channel RIMMs. These Rimms will offer from 8.5 GB/s (1066 MHz) up to 9.6 GB/s (1200 MHz). >>



(drool)



<< This is rather impressive roadmap because it is very realistic and doable, as all evolutionary steps are based on the same 16 bit DRDRAM devices DRDRAM uses today. If Rambus executes well, it will be hard for DDR SDRAM to keep up. After all, dual channel DDR SDRAM is certainly possible, but it is a rather expensive solution, so increasing the clockspeed is still the most cost effective means of increasing the overall performance of SDRAM. >>



 

xtreme2k

Diamond Member
Jun 3, 2000
3,078
0
0
RDRAM itself is a VERY GOOD technology. No question about that. Just that it is too expensive compared to DDR.
For the latency issue, it isnt really much of an issue. Faster the clock the lower the 'relative' latency to the CPU. Increase the memory clock by 1/3, its latency RELATIVE TO THE CPU decrease by 1/3.

The thing is that at 533, RDRAM does not have a 'high latency' disadvange compared to DDR anymore. But at 400MHz, it does.

I will eventually get northwood if the RDRAM is priced at a reasonable level. 512Kb L2, 533MHz FSB, High MHz.... AMD needs to rush their Palos out and at high clock also.
 

KpocAlypse

Golden Member
Jan 10, 2001
1,798
0
0
I really, really, really, need to talk to the intel rep. The old school 1.7 p4 Willy core is starting to look less and less of a bargin...




<< What do you guys think about this latency find!!!!? >>



yummy....:)
 

KDOG

Diamond Member
Oct 9, 1999
5,525
14
81
"AMD needs to rush their Palos out and at high clock also. "

I agree. Isn't AMD going to support 166Mhz FSB - 333Mhz DDR? They need to come out swingin' on the FSB/mem bandwidth issue, as well as getting their procs in laptops at a higher speed than Intels...
 

ST4RCUTTER

Platinum Member
Feb 13, 2001
2,841
0
0
Sweet! This'll just make the upcoming Northwoods that much better! Looks like things are starting to look good for Intel :D

I wonder if AMD will exercise their RAMBUS license anytime soon. I imagine it will take more than 1066 before they do that. At least not until the Hammer. Man, I can't wait till October 15th.
 

AGodspeed

Diamond Member
Jul 26, 2001
3,353
0
0
What do you guys think about this latency find!!!!?

Very very nice. 2.2 GHz Northwood and 533 FSB sounds very cool.

I just have one question. Isn't it true that PC1066 isn't going to be out until 2nd Half 2002?!?!?!? That sucks. I might as well be anticipating DDR-3!! and 10 GB/s bandwidth capable chipsets.

Oh man is October gonna be a sweet month

You're telling me! We'll know more about Northwood and Hammer by October 15th (major droolness), and the Pally and nForce will be out. :):D:):D:cool:
 

damocles

Diamond Member
Oct 9, 1999
5,105
5
81
I just have one question. Isn't it true that PC1066 isn't going to be out until 2nd Half 2002?!?!?!? That sucks. I might as well be anticipating DDR-3!! and 10 GB/s bandwidth capable chipsets.

According to that Aces Review

Samsung's DRDRAM roadmap looks promising. Right now, most of the PC800 are already able to reach 1066 MHz. At the end of this year, we should see official 1066 DRDRAM, offer 2.1 GB/s per channel.

Hopefully that's right
 

Athlon4all

Diamond Member
Jun 18, 2001
5,416
0
76
I am very suprised especially about the latency find. Truthfully, I've been very skeptical about RDRAM's future because I was very doubtful that Rambus would eaisly go to 32-bit (much less 64-bit), I was also suprised about how they're doing the 32/64-bit. Interesting way very effective. The other thing is that considering all the heat current RDRAm produces, I expected that they were going to have loads of issues just to get it to 1066, but again I was wrong. Does anybody have any info on DDR-II. It is obvious now, DDR is really going to have a rough time, even if QDR become common place, it's gonna be tough.
 

fkloster

Diamond Member
Dec 16, 1999
4,171
0
0
I have to say, as I have said for years now...High speed Serial memory interfaces will be the future.
 

Sunner

Elite Member
Oct 9, 1999
11,641
0
76
There weren't too many people that disagreed with you, most just disagreed with they way you said it, as it included quite a few that aren't too... appropriate ;)

I just hope Rambus goes under, which is not to say I hope serial memory goes under.
I dont usualy get PO'ed at companies, but in Rambus' case Im ready to make a big exception.
 

Degenerate

Platinum Member
Dec 17, 2000
2,271
0
0
I a impressed. But have 2 Q's

a) will we need a new mother board? I dont get how he jst stuck in the 533, and it ran.
b) Does P4, suppot this? as in the FSB is 400mhz (100x4), How will the p4 handle this extra 133?
 

Dogma

Member
Sep 14, 2001
60
0
0
All you people against Rimms, Rambus, and serial memory and its higher prices are complete morons. Once the facts start changing, you change...you all probably went out and bought a couple of flags after the terrorism as an act of patriotism...where were your flags before? Once Rambus memory kicks DDR's ass (oops, it already has)...you'll all say it's better...but for the longest time, you've all cried.

Jeez, 256mb of Rdram is going for around $100....jeez...what a bunch of cheepskates...so what if sdram/ddr is cheaper? Just a year ago sdram was more than rdram is right now...some idiots will never be satisfied.
 

dullard

Elite Member
May 21, 2001
25,825
4,386
126


<< I expected that they were going to have loads of issues just to get it to 1066, but again I was wrong. >>



About 3-4 months ago Rambus held a press conference saying that 1066 will be as ABC. Just stick on a few more of their chips that they already produce and some slightly different wiring and logic. No one here believed them. Can anyone explain why so many people think Rambus cannot do what it claims? Sure they may do it in an rude, unethical manner, but as far as I can tell they still should be able to achieve their roadmap without much difficulty.
 

Athlon4all

Diamond Member
Jun 18, 2001
5,416
0
76
What I meant was I thought they were gonna have heat issues, but I didn't ever hear that they said they wern't gonna have any issues.
 

AGodspeed

Diamond Member
Jul 26, 2001
3,353
0
0
Samsung's DRDRAM roadmap looks promising. Right now, most of the PC800 are already able to reach 1066 MHz. At the end of this year, we should see official 1066 DRDRAM, offer 2.1 GB/s per channel.

Then what is the Tulloch? I know that's due out by 2H 2002. Isn't Tulloch supposed to start out with PC1066?
 

WilsonTung

Senior member
Aug 25, 2001
487
0
0
Did anyone notice the i845 SDRAM latencies... 20% lower than what Anand reported for the i845 in his P4X266 review.
 

dullard

Elite Member
May 21, 2001
25,825
4,386
126


<< Did anyone notice the i845 SDRAM latencies... 20% lower than what Anand reported for the i845 in his P4X266 review. >>



I've seen that difference, but if ram makers keep increasing their bandwidth who will ever care about minor latency differences? If the memory delivers more info then the processor can handle, then the latency is essentually zero for the vast majority of memory needs. Or am I completely wrong?