Hardcore DDR2 RAM by Corsair, G.Skill, OCZ and Patriot

[pschmid]

DDR3 will reach higher clock speeds than DDR2, but it doesn't yet outperform high-end DDR2 modules. We look at some of the best high-grade DDR2 RAM available.

 

[powerbaselx]

DDR3 will reach higher clock speeds than DDR2, but it doesn't yet outperform high-end DDR2 modules. We look at some of the best high-grade DDR2 RAM available.

And what about DDR2-667 vs DDR2-800? Does it make much difference in performance for a common user or it is just like the article tests?

 

[Track]

DDR3 will reach higher clock speeds than DDR2, but it doesn't yet outperform high-end DDR2 modules. We look at some of the best high-grade DDR2 RAM available.

And what about DDR2-667 vs DDR2-800? Does it make much difference in performance for a common user or it is just like the article tests?

For the average user? God no. Not even for gamers.

 

[IcY18]

Well since DDR2-667 performs worse than DDr2-533 at stock fsb E6x00 speeds then the difference between DDR2-800 and DDR2-667 would be greater. Thats not the point of the article though. Its to test the best DDR2 memory there is. There point with the DDR3 is that it hasn't matured enough to be the highest performing memory yet.

 

[caamsa]

DDR3 will reach higher clock speeds than DDR2, but it doesn't yet outperform high-end DDR2 modules. We look at some of the best high-grade DDR2 RAM available.

And what about DDR2-667 vs DDR2-800? Does it make much difference in performance for a common user or it is just like the article tests?

Hell what about regular DDR 400? Is there that much of a difference for high end DDR?

 

[IcY18]

With DDR2 reaching much better speeds and timings like DDR2-800 3-3-3-12 or DDR2-1200, yes.

 

[fx51]

Performance gain , don’t justify price .

 

[rockowastaken]

Nice article and great introduction. Thanks for remembering us hardware newbies who read articles and think 'CL what?"

As you can see in our benchmark section, the performance results do not vary much.

As an aside, I would bet if you figure in the variance from each of your test sets you'd find no statistically significant difference between any two performance benchmarks. Certainly not enough to justify price differences between these modules.

 

[darious00777]

Just starting reading, but noticed this:

DDR is available in speeds ranging from 100 and 200 MHz (DDR200 to DDR400), while DDR2 is specified for 133, 166 and 200 MHz (DDR2-533, DDR2-667, DDR2-800). DDR3 has just been launched at 533 MHz (DDR3-1066), with 667, 800 and probably even 1 GHz on the roadmaps.

Shouldn't it look like this:

DDR is available in speeds ranging from 100 and 200 MHz (DDR200 to DDR400), while DDR2 is specified for 200, 266, 333, and 400 MHz (DDR2-400, DDR2-533, DDR2-667, DDR2-800). DDR3 has just been launched at 533 MHz (DDR3-1066), with 667, 800 and probably even 1 GHz on the roadmaps.

DDR2 didn't turn into QDR (quad data rate). Someone got their math a little mixed up.

 

[IcY18]

Should be an editor.

 

[aonhagen]

Remember whats the real purpose of DDR3.....

Go and buy ram... again.

 

[GSTe]

Testing hardcore DDR2 ram and didn't include Team or especially Ballistix! :lol: :lol: :lol: genius :lol: :lol:

 

[david__t]

Yet again the issue of the ratio is skipped over - although it did get a mention. This really needs looking in to as everybody accepted during the DDR period that synchronous settings were better until you reached such high memory speeds that you started to gain any benefit. This issue needs re-addressing for DDR2 & DDR3 - especially in the light of the upcoming FSB1333 CPUs.
I still believe that a small FSB overclock to match your memory speed keeping the 1:1 ratio is far better (and easier to acheive with little extra cooling expense) than trying to reach massive memory speeds - and to a certain extent this article backs that up by saying many times that the expense of super high speed memory modules is not really worth it for 99% of people.
PLEASE THG - a comprehensive guide to ratios.

 

[darklife41]

DDR3 will reach higher clock speeds than DDR2, but it doesn't yet outperform high-end DDR2 modules. We look at some of the best high-grade DDR2 RAM available.

And what about DDR2-667 vs DDR2-800? Does it make much difference in performance for a common user or it is just like the article tests?

For the average user? God no. Not even for gamers.


I think the 1st and last paragraph of the article answered this very well. The entire point of this comparison is not for the average user, nor those on a budget. Many applications see a sizeable increase in speed with DDR2-800 and up, and many applications don't. It really comes down to what you use your PC for.

I also thought any hard core gamer worth their salt would be heavily into overclocking, and in that situation DDR2-800 and up are definitely worth considering. You won't get much of an overclock with 533, and still pretty limited with 667.

I'd guess Crucial Ballistix PC2-6400 would give any of the listed Dimms a run for their money. It will definitely beat the G-Skill PC2-6400 and tighten up the timings to compare with the Patriot and Corsair. I have Ballistix PC2-6400 running 1-1 with an E6400@3.27GHz with 4-3-3-8 timings. My constraint is the motherboard. (CPUZ lists: Rated FSB 1636MHz, Core Speed 3272, Bus Speed 409MHz)

 

[bfellow]

Actually you can go here to see a difference with DDR2 800 all the way up to 1074 based on different brands and timings.
It shows WinRAR is tremendously affected (220kb/s difference b/w top and bottom)

G.Skill HK Memory Review

According to this site the memory I have overclocks to 940 mhz on stock voltage and 4-4-4-12 timings.

 

[darklife41]

I don't know what memory you have. I just know that this article states they could not tighten up the timings on the G-Skill they were reviewing no matter what they tried. Crucial Ballistix PC2-6400 will allow timings of 4-3-3-8 up to around 430MHz, so is faster than the G-Skill memory up to that speed.

After a BIOS update, I was able to OC my E6400 to 3.6GHz with the Crucial PC2-6400 running 1:1, 4-4-4-12, @ 920MHz. I didn't try setting dividers to see how far the memory will go because I'm only interested in running 1:1 for overall performance. Apparently 3.6GHz is the max for this particular E6400 on air. It would boot at higher settings (463 FSB) and passed super-pi v1.5 at 1MB without a problem, but crashed during 3dmark benchies.

CPUZ shows the SPD at 4-4-4-12@400MHz and 5-5-5-15@500MHz. So I still expect the Ballistix to out-perform the G-Skill.

 

[darklife41]

Actually, I just upped the CPU to 465GHz, which is 930MHz on the memory and ran SuperPi. The memory is still stable, but the CPU is well past stable. I got screenshots of the Crucial at 840GHz with 4-3-3-8 timings and at 930MHz at 4-4-4-12 timings, both after running SuperPi (1MB). I have some errands to run and then will get links posted to the screenshots.

The point is that Crucial Ballistix PC2-6400 is better than the G-Skill reviewed.

 

[darklife41]

Here's a link to the screenshots.

http://www.ultramaxcc.com.au/crucial6400.htm

 

[billyhoyle]

I think they must have gotten a bad batch of G.Skill. I have mine running at 916 MHz, 4-4-3-7 2.1v, no errors in 12 hours of memtest and prime95.

No screenshots, sorry.

 

[darklife41]

I still say any review about high end RAM has to include Crucial to be worth reading. And I'm not impressed with this particular model of G Skill.

BillyHoyle,
The numbers in your sig don't seem to add up, but since I don't have a E6320 on an overclockable MB at the moment I don't know for sure. Your memory is running 1:1 at 920MHz with your CPU at 3.2GHz?

Here's my CPU at the same speed, and while your vid card is better, mine is beating you by 864pts on 3dmark05? Hmmm. (12,229)
http://service.futuremark.com/compare?3dm05=3190764

This one just shows that the CPU speed doesn't make that much of a difference in score. (.3GHz faster = 100pts with everything else the same 12,330)
http://service.futuremark.com/compare?3dm05=3192506

And this one was run at 3.2GHz with generic DDR2-6400 at 5-5-5-15 while I was waiting on my Crucial. (11,916)
http://service.futuremark.com/compare?3dm05=2529663

Now I don't know for sure how much difference memory makes in these scores. But generic RAM with a 7900 beating out G Skill with a 7950 has to be worth noting.

 

[billyhoyle]

The 7900gtx is a higher end card than a 7950gt - hence the higher 3dmark scores. The 7950gt is basically a 7900gt with HDCP support. I think you may have thought I have the 7950gx2, which is a monster.

The multi on the e6320 is 7x, as opposed to the e6400 is 8x. My FSB is 458 MHz with RAM at 1:1 giving 916 MHz, and a cpu clock of 3206 MHz.

With my CPU at stock (1.86) I get about 10k 3dmarks, and at 3.2 GHz, get about 11k. I've run it with timings as they are now, and at 5-5-5-15, but don't notice a significant difference in scores.

I agree with you that crucial should have been included for review. I've never used it personally, but have heard only good things.

Sorry I don't have any links/pics, don't have any handy.

 

[darklife41]

No worries, and I do appreciate the explanation.

 

[KvnTrny]

All i want to know is how to get my 2x1GB Patriot DDR2 800 to run. Im replacing what was 512 MB and the monitor doesn't turn on. Prolly wrong topic but I've been struggling with this for over a month haha

 

[Blackguard83]

While reading the article and being an average user, who never even attempted to build a system or overclock something.. anything! Two words have crossed my mind all the time. RIP OFF!

On a more serious note, yes, it was stated that the whole point of this article is to compare high end hardware for enthusiasts and it is not orientated to someone like me so that's probably explains why the author wasn't so one dimensional in his conclusions like me.