*The Official Tech-Forums 3DMark06/03 Rankings*

[gurusan]

Re: The Tech Forums Official 3DMark06/05/03 Top List

I'm thinking its the 1.5gb of ram, if i recall correctly Oblivion is a really GPU limited game, a X2 chip at 2.9ghz should be sufficient.

 

[Apokalipse]

Re: The Tech Forums Official 3DMark06/05/03 Top List

Think you might be mistaken there.

not to try and start an argument here, but I don't think so....
I've been able to get >8K SM 3.0 score and >6.5k SM 2.0 score (Crysis hotfix drivers)
CPU score was >2600

 

[gurusan]

Re: The Tech Forums Official 3DMark06/05/03 Top List

alright well I just showed you proof that CPU clockspeed has a direct impact on SM2 and SM3 scores with a 38xx crossfire setup, and there's a lot more on ORB.

 

[Apokalipse]

Re: The Tech Forums Official 3DMark06/05/03 Top List

alright well I just showed you proof that [a significant increase in] CPU clockspeed has a direct impact on SM2 and SM3 scores with a 38xx crossfire setup, and there's a lot more on ORB.

You get a much bigger difference in your score with a quad over a dual.

 

[gurusan]

Re: The Tech Forums Official 3DMark06/05/03 Top List

In overall score yeah because of the CPU score, but not in SM2 and SM3 scores. Especially because the phenoms don't clock nearly as well as the X2 chips. Just look on orb and the Phenom 3dmark06 sm2 and sm3 scores are not very impressive in comparison to higher clocked scores from the X2.

 

[Apokalipse]

Re: The Tech Forums Official 3DMark06/05/03 Top List

In overall score yeah because of the CPU score, but not in SM2 and SM3 scores.

But that's kinda my point.
the SM 2.0 and 3.0 scores are much more GPU bound; but even so, my 6000+ is not really a bottleneck to the graphics cards. (the GPU tests will only utilise one CPU core anyway)
but the CPU test does make the overall score lower, because it's not a quad.

Especially because the phenoms don't clock nearly as well as the X2 chips. Just look on orb and the Phenom 3dmark06 sm2 and sm3 scores are not very impressive in comparison to higher clocked scores from the X2.

Not many people are actually buying Phenoms to overclock them

but B3 stepping Phenoms can get >3GHZ, and at those clock speeds, perform better per core than X2's. Plus, they're quad.

 

[gurusan]

Re: The Tech Forums Official 3DMark06/05/03 Top List

...yet the fact remains that the GPU scores are lower, which was my original point (switching to a quad will not relieve any bottlenecking on your CF setup in the 1280x1024 3dmark06 test, in fact it's more likely to do the opposite). Unless you manage to get the Phenom to 3.4+ ghz.....

Not saying the phenom wouldn't be a good investment, I'm only talking about this specific 3dmark06 test.

 

[Apokalipse]

Re: The Tech Forums Official 3DMark06/05/03 Top List

...yet the fact remains that the GPU scores are lower

They're not really much lower on my 6000+
That was my point.

switching to a quad will not relieve any bottlenecking on your CF setup in the 1280x1024 3dmark06 test

See, I don't agree that they are really bottlenecked. Not the GPU scores, anyway.

Unless you manage to get the Phenom to 3.4+ ghz.....

a phenom at 2.8GHZ will perform about the same per core as my 6000+ at 3.4

In fact, I'm planning to buy Dainius's Phenom 9850 BE, which he got to 3GHZ on air (stock), using an Asus Crosshair (which is not ideal to use for a Phenom)

I have an MSI K9A2 Platinum (790FX AM2+ board), and watercooling.

 

[gurusan]

Re: The Tech Forums Official 3DMark06/05/03 Top List

hmm....just look on ORB, there's a decent difference in SM2 and SM3 scores between 3.5ghz x2 scores and 3.0ghz Phenom scores using 3870 cf setups. I'm talking 800-1000 pts which is pretty large when talking about individual sm2 and sm3 scores.

Also I showed you the test I had run myself on an intel system. 20% more GPU performance was gained on an identical setup, the only thing different was 550mhz and 2mb cache on the cpu.

If that's not evidence that clockspeed has a large impact on crossfire bottlenecking, then I don't know what is.

 

[Apokalipse]

Re: The Tech Forums Official 3DMark06/05/03 Top List

I've checked ORB, there isn't a real difference between X2 3.4-3.5GHZ systems and Phenom 3GHZ systems using crossfired 3870's

Also I showed you the test I had run myself on an intel system. 20% more GPU performance was gained on an identical setup, the only thing different was 550mhz and 2mb cache on the cpu.

that is a significant difference.