• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

whats the best buy? ddr2 800 or 1066

mentalcrisis00

mentalcrisis00

Senior member
Wondering which would give me the better performance

DDR2 800 with 4-4-4-12 timings or

DDR2 1066 with 5-5-5-15 timings

also whats the cpu of choice these days for gaming? core 2 duo, core 2 quad, or AMD phenom?
 
performance would be identical. the only reason to get DDR2 1066 is for overclocking via the FSB, unless you actually need higher memory bandwidth. CPU of choice is a core 2 (duo or quad depends on what you're using it for), especially if you plan on overclocking any
 
For gaming, my understanding is the dual cores are generally superior to the quads, because they run at higher clock speeds than quads, and almost all games do not take advantage of all 4 cores in a quad.

I'm primarily a gamer, but I have a Q6600. I just enjoy the satisfaction of knowing I have four cores. 😉

I'm finding that even at stock speeds my game performance is phenomenal (standard 1600x1200, 70+ fps, AA + AF on), and I don't know if I'd even notice a difference between a Q6600 @ 2.4GHz and an E8400 @ 3.0GHz.

Edit: I came to these forums with the same question - 800 or 1066? Forum searches showed contradictory results, so I'm still not clear on it. Some said the Intel CPUs love the higher bandwidth, others said that bandwidth would never be achieved, so only get 1066 if you really need it for OCing. Eventually I went with these Mushkin sticks: 2x2GB 5-4-4-12 DDR2 800.
http://www.newegg.com/Product/...x?Item=N82E16820146731
 
On Core2 latencies are not as critical as on past platforms. The more CPU and memory speed you can get, the better. So, if CAS5 nets a higher speed on memory or CPU, it is typically better performing than a slower speed at CAS4. Some will argue this in a benchmark context but, real world, speed FTW.
 
Originally posted by: DSF
In the real world it makes essentially no difference in 99% of applications.
Click to expand...
Acutally in my last round of testing when 975X chipsets came out, it averages 2%-5% over a variety of apps and tasks. That's using a memory divider and running the same CPU speed to the CPU would not skew the results. At that time, the premium for 8500 vs 6400 turned a lot of people off for the potential gains. But, these days, IMO 8500 is the way to go.

 
Originally posted by: Yellowbeard
Originally posted by: DSF
In the real world it makes essentially no difference in 99% of applications.
Click to expand...
Acutally in my last round of testing when 975X chipsets came out, it averages 2%-5% over a variety of apps and tasks. That's using a memory divider and running the same CPU speed to the CPU would not skew the results. At that time, the premium for 8500 vs 6400 turned a lot of people off for the potential gains. But, these days, IMO 8500 is the way to go.
Click to expand...

I don't see the difference between 50 and 52 FPS as significant. Maybe that's just me.
 
Originally posted by: DSF
Originally posted by: Yellowbeard
Originally posted by: DSF
In the real world it makes essentially no difference in 99% of applications.
Click to expand...
Acutally in my last round of testing when 975X chipsets came out, it averages 2%-5% over a variety of apps and tasks. That's using a memory divider and running the same CPU speed to the CPU would not skew the results. At that time, the premium for 8500 vs 6400 turned a lot of people off for the potential gains. But, these days, IMO 8500 is the way to go.
Click to expand...

I don't see the difference between 50 and 52 FPS as significant. Maybe that's just me.
Click to expand...

I did not claim it was significant per se. I just threw some numbers out there from actual real world testing, not synthetic benchmarks. However, I guess a 5% increase to some would be significant.

And, keep in mind that memory bandwidth rarely ever affects gaming MAX FPS anyway. My focus was on video rendering, graphics rendering, compression, etc etc.
 
Originally posted by: mentalcrisis00
Wondering which would give me the better performance

DDR2 800 with 4-4-4-12 timings or

DDR2 1066 with 5-5-5-15 timings
Click to expand...
If you can find a good deal on DDR2-1066, run it at DDR2-800 @ 4-4-4-12 with a tad less voltage. Most 'enthusiast' module companies are just overvolting chips to get to the next highest speed bin, anyway. So you might see modules like this:

DDR2-800 4-4-4-12 @ 1.8V

DDR2-1066 5-5-5-15 @ 2.0V

HELLO! Yes, they are actually taking DDR2-800 IC that tolerate excessive voltage to make DDR2-1066 modules. Problem is, they often push it almost to the stable limit, which is why you get so much variability from motherboard to motherboard.

If you take it back to where it supposed to be, it restores some of the operating and electrical margin, and magically the modules work fine on a whole lot of motherboards they didn't work so well on at the higher speed/voltage.
 
Originally posted by: mentalcrisis00
Wondering which would give me the better performance

DDR2 800 with 4-4-4-12 timings or

DDR2 1066 with 5-5-5-15 timings
Click to expand...

Originally posted by: faxon
performance would be identical
Click to expand...

Nope, look here: http://www.overclock.net/intel...get-2.html#post4770288

Intel processors love bandwidth and more MHz make them perform better.
Everest, SuperPi, and Sandra will all show improvement with greater bandwith.
As for real world improvement, that depends on your system and applications.
 
Originally posted by: 18 Is Number 1
Nope, look here: http://www.overclock.net/intel...get-2.html#post4770288

Intel processors love bandwidth and more MHz make them perform better.
Everest, SuperPi, and Sandra will all show improvement with greater bandwith.
As for real world improvement, that depends on your system and applications.
Click to expand...
Higher memory frequency always increases raw memory bandwidth, which is what synthetic memory bandwidth tests are measuring. It could hardly be otherwise, whether its Intel, AMD, PPC, ARM, or MIPS.

That is NOT synonymous with "better performance". At best, *some* applications might show 3% increased performance between DDR2-800 and DDR2-1066, most others won't even be 2%.
 
For some games, a few extra fps is a big deal.

DDR2-1066 5-5-5 will beat DDR2-800 4-4-4 always unless you are running nForce 1:1 synced with 1T.

But overall, the money is usually better spend on a better CPU/GPU, etc., as other than synthetics, the real world performance improvements of using higher end RAM are far too small to worry about.
 
Originally posted by: n7
For some games, a few extra fps is a big deal.
Click to expand...
Except when the few extra fps amounts to 2% ~ 3%, which would mean:

20 --> 20.6

50 --> 51.5

100 --> 103

Sure, you might see the difference between 10 and 13 fps, but that would be a 30% increase in performance, not 3%. You'd be lucky to find a 30% increase in application performance between DDR2-667 and DDR2-1066.
 
Originally posted by: 18 Is Number 1
I'm sorry, I must have missed something. An improvement of 2-3% isn't better performance?
Click to expand...

Would you notice the difference between 50 and 51.5 or 60 and 62 FPS? 10 and 10.3? I wouldn't.

Now, if you encode thousands of videos, it might matter to you that they finish encoding in 58 minutes instead of 60. For most people, the difference is meaningless.
 
Originally posted by: mentalcrisis00
Wondering which would give me the better performance

DDR2 800 with 4-4-4-12 timings or

DDR2 1066 with 5-5-5-15 timings

also whats the cpu of choice these days for gaming? core 2 duo, core 2 quad, or AMD phenom?
Click to expand...

Relating this back to the OPs question, DDR2 1066 will offer better performance than DDR2 800.

Intel Core 2 8*** series CPUs are the weapon of choice.
 
Xbitlabs comparison: DDR2-800 up to DDR3-1333 & CAS 3 to 9

In gaming benchmarks, they saw minimal differences. Increased bandwidth beats lower cas ratings in every case. However...the differences are laughably small and certainly don't justify the higher prices the RAM producers ask for faster memory. I mean, $70 1066 versus $50 800 is a 40% increase in cost for maybe a 1-2% improvement in performance?

I personally will stick to my CAS5 DDR2-800, thanks very much.
 
DDR1066 memory really shines when compared to DDR800 that cannot OC well enough to keep up with a CPU going past a 400FSB. And, as already noted, MAX FPS in games rarely shows any improvement from sheer memory bandwidth. But, we really don't know this users needs so it may or may not be gaming.
 
Originally posted by: VirtualLarry
SeventeenorBust loves memory bandwidth.
Click to expand...

Actually, almost all DC projects seem to love bandwidth, especially the ones that either fold proteins or work with extremely large numbers. Back when I used to fold proteins 24/7, I got a huge gain when I upgraded my RAM to BH-6 @ 2-2-2-6, from RAM that could only do 2.5-3-3-7 (or was it 8?). The two sets were running at the same speed in Mhz, but I was getting something in the neighborhood of 10% more performance from the lower latency RAM. With that said, if you aren't doing scientific work or overclocking a low multiplier processor, save your money.
 
Originally posted by: 18 Is Number 1
Originally posted by: Engineer
I recommend the cheap 4-4-4-12 stuff! 😛
Click to expand...

That's good...buy $25 RAM that you'll spend $50 in shipping to RMA over and over again.
Click to expand...

YMMV. There was a problem with the Ballistix stuff at one time but it has since been revised. As with any product, you can get a bad batch once and awhile. I've only had one stick of ram die in 24 years of PC tinkering so I always take the "cheap" chance! 😛

Oh, and why would you spend $50 to ship over and over? Just buy new ram for $20 a few weeks later! 😀
 
Agreed. Crucial has since lowered the max vDIMM to 2.0 from 2.2v. That said, even the old PC2-6400 never needed more than 2.05v for 533 FSB speeds -- and that was with 4 slots filled too. I now use 1.90v with 4 slots filled at 480mhz with this new set.

I have a 2x2gb set en route so I'll let you know how that set works out, but I think the point is, despite the rated voltage on ANY set, you should use the absolute minimum vDIMM you can get away with.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top