DDR 2100 price vs RDRAM price

[juin]

I have look at pricewacht for DDR 2100 256 mgbytes Cas 2 from micron cost 51$ to 56$ vs RDRAM PC 800 256 mgbytes from samsung 74$ to 75$. The marge is no more big.

So in a 512 RAM syteme there is a 42$ marge.

Let do the price/performance

DDR 2100 have something like 15% less lantency.RDRAM offer 3.2 gb/s vs 2.1 gb/s So a 51% increase.51-15=36%

So 51$ x 136%= around 68$

So at the end 68$ vs 74$.

http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670

 

[xxsk8er101xx]

remember with rdram you have to buy two sticks

<A HREF="http://www.anandtech.com/mysystemrig.html?id=9933" target="_new"> My Rig </A>

 

[AMD_Man]

You can't compare the performance of RDRAM and DDR in that fashion. They are two extremely different technologies!! In most cases, the lower latency of DDR RAM more than compensates for the higher bandwidth of RDRAM. You can't directly compare the bandwidth of RDRAM and DDR RAM because DDR is far more efficient than RDRAM.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[Matisaro]

Another half assed comparison by juin, totally neglecting how things really work.


AMD_man summed up all my complaints with your logic juin, try again when prices fluctuate close once more.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

 

[girish]

bandwidth figures dont matter alone.
right, the latency somewhat compensates for the lower bandwidth and delivers within 10% performance even with 50% less bandwidth. see the P4X266 v/s i850 comparision on anandtech.

there is one more reason for such small performance difference. any processor does not access the memory all the time. its has to do the I/O on the same bus all the time. traditionally the x86 processors have the I/O devices mapped on the same bus but different address space. you can safely assume thaat the processor spends about 30% of its time accessing the I/O devices, and that time it isnt accessing the memory. In short, the CPU itself isnt accessing the memory at 3.2GB/sec!

Chipsets these days have buffer for memory so the actual bandwidth of the external memory does not matter much as long as the CPU is fed properly. Modern chipsets can decouple the CPU FSB and the memory bus to such an extent that totallly different technologies can be used. Like the i845 chipset works on the P4 side at its native bus of 400 MHz while controlling a simple PC100/PC133 memory on the other! This is justified by the the difference in performance of P4 with RDRAM and SDRAM, when the bandwidth of SDRAM is just 33% of RDRAM! Memory intensive applications like games and graphics lose severely with SDRAM thanks to its lower bandwidth.

Now the FSB does not count, what counts is the bandwidth! That is why the MHz rating is gone and bandwidth ratings are on. Now you dont call a memory as PC133, PC266 etc, you call it PC800, PC1600 or PC2100!

girish

<font color=red>No system is fool-proof. Fools are Ingenious!</font color=red>

 

[juin]

see the P4X266 v/s i850 comparision on anandtech.

All DDR benchmark i have found are base with Abit th7-2 wich is slower that P4t.All AMD systeme have allwayse the best mobo.If there is a P4T-E or P4T they use a old P4 1.4 or 1.5 wich dont really need bandwith.Enable turbo RIMM with a P4T and look DDR fall back.

http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670

 

[juin]

anandtech

I dont really trust any P4 benchmark from anadtech simply they suck.

http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670

 

[Ganache]

It's true that now RIMMs has better bandwith but it's because the i850 has a dual channel for memory. That's why they have 3.2G
Look at the memory bandwith of a nForce chipset. It's 4.2G/s because of the dual channel of DDR.

Now it's a more fair comparison.

The only way we will see what's better if when we compare DDR and RDRAM with the same CPU.

Athlon can't use all the bandwith because of the "slow" fsb compared to the P4.(266 vs 400 sure makes a difference )

lets put a p4 in a nForce (maybe someday) and im sure RAMBUS goes down. That why Intel and Rambus don't want nvidia to aquire liscence for p4. They will prove that rambus is poop.

 

[juin]

What tell you that NFORCE dual memory controller really work they can just write and will enable it later.

http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670

 

[Guest]

I see 128 Mb RDRAM at pricewatch for $33
128 Mb PC2100 DDR $18 (granted CAS 2.5, but why compare to CAS 2 ?) Thats nearly half the price.

Either way, its a good thing for P4 users RDRAM is becoming that affordable.. WHo knows.. maybe I would want an Athlon + RDRAM system for SiS one day.. if AMD brings out a 200 Mhz DDR fsb cpu at least..

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Kelledin]

Just out of curiosity, juin, can you list for us all the sites that (in your opinion) suck?

Kelledin
[dave@discovery ~] kill -9 1
init: Just what do you think you're doing, Dave?

 

[Stiffler]

Hmm I only have one stick, as long as you use a continuity module in the other slot its fine !


Tim

I am Homer of Borg ! Prepare to be... MMMhhhhh Doughnuts

 

[Matisaro]

In a p4 dual rambus system, having 1 rimm halves your bandwidth and ups the latency, and generally is not good I believe.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

 

[Stiffler]

so would I see improvement if I bought another ram module ?

Tim

I am Homer of Borg ! Prepare to be... MMMhhhhh Doughnuts

 

[Matisaro]

Nope, you have a p3, only a p4 would benifit from the dual rambus, the p3 actually would suffer due to the increased latency on the ram with more than 1 module.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

 

[xxsk8er101xx]

maybe but you reduce the memory bandwidth in half and reduce performence.

it uses dual channel technology. So you need two sticks to have the needed bandwidth. right?

<A HREF="http://www.anandtech.com/mysystemrig.html?id=9933" target="_new"> My Rig </A>

 

[Crashman]

P3's use CRIMMs, P4's use matched pairs.

What's the frequency, Kenneth?

 

[girish]

The <A HREF="http://www.anandtech.com/showdoc.html?i=1518" target="_new">Anandtech P4X266 v/s i850 review</A> used a reference board for P4X266, a reference board is just a engineering sample which usually performs lesser than the actual product. And they used a Abit BH7 board to get the SDRAM benchmarks, and used the MSI 850Pro2 for the 850 and not the Abit one.

And whats the cry about AMD getting better boards and Intel getting the bad ones??? Face it, you are disappointed because the results are not what you like.

girish

<font color=red>No system is fool-proof. Fools are Ingenious!</font color=red>

 

[juin]

1 i dont really care about p4x266 because there is allready bnchmark of P4X266A who is a bit faster that I845-D in fact P4x266a is equal or faster that SIS645/DDR333.

2 MSI pro2 hoo way better that a Abit Th7-2 socket 478.I want a P4 2.0 with a P4t-e benchmark.Why you wonder that toms allwayse have better score with P4 that anadtech/acehardware ..... (kelledin that have nothing to do with your benchmark from ace hardware.)

http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670

 

[juin]

I can really recall all site that i dont like but i can tell you my favorite.

Tomshardware reference benchmark
Hardware.fr Benchmark score very similiar to tomshardware.News there is all the rumor and official release.They are also more realistic in there bechmark.
XBITLAB they have also a lot of interessing news.
Firingsquand (there is a error in the word) the made a lot of game benchmark score also similiar to tomshardware.



http://gamershq.madonion.com/products/orb/?publish_compare.shtml?&project_type=6&dprid=2108670