512 DDRAM vs 256 RDRAM

[Guest]

I understand that 256 (half of 512) RDRAM is BETTER than 512 DDRAM for a P4?? is this true? how much better?

 

[FatBurger]

Why would it matter? PC800 and PC2100 are pretty much the same price now, just get 512MB of RDDRAM.

But to answer your question: 256MB of RDRAM would be better under nominal usage, but if you have a lot going on at once, 512MB of DDR would be better.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft

 

[Guest]

you didn't answer my question. I didn't say i don't have the money to buy 512 (which I don't) I asked that in order to know how efficient is RDRAM compared to DDRAM for a P4. Can someone answer my question?

 

[FatBurger]

RDRAM provides around 2.5GB/s of bandwidth at stock clock speeds, DDR provides about 1.8GB/s (averaging things out here).

My OCed P4 system gets around 3.24GB/s in Sandra, someone else will be able to tell you about a decently OCed DDR system.


EDIT:
I was looking through my benchmark screenshots just now, and realized that with a 155MHz FSB and RAM at 3/4 I get 3.24GB/s, not 3.3GB/s. 133MHz with full speed RAM gives me 3.3GB/s. Sorry about the error.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft<P ID="edit"><FONT SIZE=-1><EM>Edited by FatBurger on 04/01/02 07:37 PM.</EM></FONT></P>

 

[Harisahmed]

All of the performance tests that compare the memory performace between RDRAM and DDR have been using the same amount of memory, what kind of a difference do you get by increasing the amount of DDR? Would a system with 512MB DDR perform better than one with 256MB RDRAM if they were in an equivalent system?

 

[FatBurger]

As I said before, it depends on what kind of tests you're running. Most benchmarks test speed only (in fact, they do better with less RAM sometimes).

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft

 

[Harisahmed]

Here are some questions: Does anyone know someone who took their existing system and got new benchmarks after increasing their memory? If so, what % increase was noticed in performance? Since we already have comparisons between similar systems, couldn't we then apply the % increase to the scores for a DDR system (taken with a grain of salt) and recompare them to the values for the RDRAM system? Although not an exact measurement, this would show approximately what the difference could be, or am I completely off base?

 

[FatBurger]

It depends heavily on your application, such a test would be widely mis-read.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft

 

[camieabz]

I understand that 256 (half of 512) RDRAM is BETTER than 512 DDRAM for a P4??

Why?

'Better' is a very unrealistic term for this comparison.

512MB of SDRAM may be 'better' then 256MB of the others, <b>if</b> you constantly use more than 256MB of RAM! Granted, the SDRAM will lag a lot in comparison, but it still goes faster than your swapfile!!!

<b><font color=blue>~ Whew! Finished...Now all I need is a Cyrix badge ~ </font color=blue> </b>

 

[Harisahmed]

I never claimed that this would be an accurate test, it is only a quick and dirty way to see if having more DDR memory would offset the higher memory bandwidth of RDRAM. As long as the tester specifically mentions that it isn't a completely accurate/scientific test, I don't see a problem with someone posting some initial results for a series of tests similar to what is used during the reviews here on THG. What would be the problem with showing a %increase in scores of similar systems, to those of us that can actually read and understand the test methodology and results? What if the tester didn't show a %increase, but just told us if it tips the scales in favor of the DDR system?

I'd still like to know, because depending on the prices for memory, I've been able to easily configure an AMD system using twice the amount of memory of a comparable Intel system.

 

[FatBurger]

I didn't mean that it wouldn't be accurate, I meant that it would probably be misunderstood by many people.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft

 

[Harisahmed]

Sorry about the misunderstanding. I didn't think there would be that many people that would misinterpret the results, because it looks like most of the people on this forum have enough technical knowledge and skill to build and/or OC their PCs.

 

[FatBurger]

On this forum, perhaps not. But for all the people who'd read such an article, it would be a different story.

<font color=blue>If you don't buy Windows, then the terrorists have already won!</font color=blue> - Microsoft