C2D & gaming

[Sparrow_69]

Does gaming benefit from a conroe cpu using 667mhz ram vs. 800mhz ram if the cpu is clocked at the same frequency (only memory divider changed to achieve these different speeds) ?

 

[Poncho]

Yes... but not noticably so. You'll benchmark higher... but I would doubt you'll see a difference. If you aren't overclocking just stick with whatever is cheapest. Though I would get 667 at least. If you'll notice I'm running 667mhz with my X6800 and it runs like a champ.

 

[chrisf6969]

you'll get higher sisoft memory scores & thats about it. Any other benchmarks would be less than 1% difference. At least from the overclocked results that I've tested. MAYBE at default FSB 266, more dram bandwidth MIGHT help, but I doubt it.


I've tested this out with multiple combinations, and they all performed about the same: (dram ratios's quoted as Gigabyte calls them, but 2:1 is actually = 1:1 )

All of these speeds with various timings like 4-4-4-12 & 5-5-5-15

9x333 = 3Ghz with 2.5 dram ratio
9x333 = 3Ghz with 3.0 dram ratio (ddr2 1000!!)
8x375 = 3Ghz with 2.0 dram ratio
8x375 = 3Ghz with 2.5 dram ratio
7x429 = 3Ghz with 2.0 dram ratio

sisoft memory scores ranged from 6000-7000mb/s depending on the Cas, FSB & dram ratio (17% change), but the actually performance was only about 0.6% at best between the best & worse performance!

 

[Sparrow_69]

When you say 0.6% difference, are you talking about the difference when gaming?

 

[(cf)Eclipse]

When you say 0.6% difference, are you talking about the difference when gaming?

most likely. ram speed traditionally has very little effect on performance in my testing with a few different cores (no conroe yet though, but i don't see why it would be excluded, especially with it's awesome prefetching)

 

[Sparrow_69]

So then for someone who is looking to oc a bit a 6400, am I better off getting pc6400 4-5-4-15 (2.1V) or pc5400 4-4-4-12 (1,9V) OCZ if price is the same?

 

[chrisf6969]

When you say 0.6% difference, are you talking about the difference when gaming?

I'm talking about the difference on the 3 benchmarks that I ran.
Super PI 1M, 3dMark 2001, 3dMark 2006

I've found that 3dMark 2001 is normally quite responsive to changes in CPU & memory speeds.

Super Pi is mainly CPU dependent but also memory bandwidth FSB can make a difference

3dMark 2006 b/c I figured it would representative of newer games, IE: videocards make all the difference, and CPU plays a small portion.

You asked for changing the dram ratio at the same FSB which makes even less of a difference than what I did (drop the multiplier and cranked up the FSB).

 

[(cf)Eclipse]

So then for someone who is looking to oc a bit a 6400, am I better off getting pc6400 4-5-4-15 (2.1V) or pc5400 4-4-4-12 (1,9V) OCZ if price is the same?

assuming price is the same, the pc6400.

 

[chrisf6969]

assuming price is the same, the pc6400.

QFT - 6400 all the way

Cas and bandwidth are not very important to Core 2 processors.

But IF YOU'RE OVERCLOCKING..... you're going to need fast ram to reach the fast FSB.

What you want is the fastest possible CPU clock speed and for that you need to increase your FSB, and since 965's lowest dram ratio is 1:1 you usually need at least DDR2 800 to max out your processor.

As you crank up your CPU speed, the increase in performance is nearly linear with your clock speed. So CPU clock speed is bar none the most important thing, and for Core2 you need fast FSB's to really max them out.

6300 (multiplier = 7x).... you need 400FSB (DDR2 800) just for a 2.8Ghz overclock!
6400 (multiplier = 8x).... you need 400FSB (DDR2 800) just for a 3.2Ghz overclock!
and many people are reaching higher.... IF their motherboard or ram isn't holding them back.