Help with EVGA P67 FTW Graphics Performance

[loddie]

Old Rig:
Asus P6t Deluxe
Intel Core I7 920 - Overclocked to 4.2ghz
Patriot Viper 1600mhz c8 - Overclocked to 1700mhz
2 Evga Gtx 465 (SLI)

Just as an example benchmark test on Heaven DX11 Benchmark 2.5

Average FPS = 50
Max FPS = 117
Score = 1527

New Setup:
Evga P67 FTW
Intel Core I7 2700K - Overclocked to 5.2ghz
Corsair Vengeance 1866mhz - stock clocked
2 Evga GTX 465 (SLI)

Just as an example benchmark test on Heaven DX11 Benchmark 2.5

Average FPS = 41
Max FPS = 99
Score = 1092

After an hour on the phone with evga tech support they have no idea and said that if i wanted to i could rma the board. I have only had the board for 2 days now does anyone have any idea of the problem and how to fix it or is it just a problem with the board and i should rma it. Has anyone else had low fps using the EVGA P67 FTW board? Would i be better off to send the board back to tiger direct for a refund and got to say the Asus board? I really love everything about this board other then this FPS issue. I have tried benchmarking and playing games and my frame rate is low on everything and my scores are low. This set up seems a lot faster when doing stuff in windows then my 920 did but i didnt just spend all this money to have slower gaming and benchmarking performance even thought i can over clock stable to 5.2ghz

Someone please help!!!! Thanks in Advance

[loddie]

Just some cpu bragging rights lol. 5.5ghz not stable enough to benchmark but cool to see it go that high

[cadaveca]

Did you re-install OS fresh for the new build?

Also, are you sure there is no throttling on CPU?

[radrok]

Have you checked if your pci slots are giving 8x/8x to your graphics cards?
It might be that one gpu has 16x and the other has 4x

[loddie]

Originally Posted by cadaveca

Did you re-install OS fresh for the new build?

Also, are you sure there is no throttling on CPU?

yes on both

clean install and speed stepping is off

[loddie]

Originally Posted by radrok

Have you checked if your pci slots are giving 8x/8x to your graphics cards?
It might be that one gpu has 16x and the other has 4x

one says pci-e x16 @16 and the other says pci-e x16 @8

[sneekypeet]

I dont want to put words in daves mouth here, but SpeedStep has nothing to do with 1.55V to the CPU overheating it and the CPU cutting things back "throttling" as you overheat that CPU

running 16x X 16x vs. 8x X 8x is going to make like 3% change at most, you have like a 20% loss.

[loddie]

Originally Posted by sneekypeet

I dont want to put words in daves mouth here, but SpeedStep has nothing to do with 1.55V to the CPU overheating it and the CPU cutting things back "throttling" as you overheat that CPU

Im not running vcore of 1.55 daily that was just to take a screen shot of 5.5ghz i run daily 5.2ghz at vcore 1.494 im liquid cooled and temps stay 65 or below C

[sneekypeet]

1.494V isnt much better. Keep an eye on what the CPU cores are doing as you run the bench, see if it is throttling Unless this is under "exotic" cooling, that is too much voltage imho for daily running.

[radrok]

Originally Posted by sneekypeet

I dont want to put words in daves mouth here, but SpeedStep has nothing to do with 1.55V to the CPU overheating it and the CPU cutting things back "throttling" as you overheat that CPU

running 16x X 16x vs. 8x X 8x is going to make like 3% change at most, you have like a 20% loss.

But running 16x/4x instead of 8x/8x or 16x/8x wouldn't be a major bottleneck for the second gpu?
That was what I meant, sorry if I explained badly

Clearly this isn't the issue anyway because gpu-z reports 16x/8x

[loddie]

Originally Posted by loddie

Im not running vcore of 1.55 daily that was just to take a screen shot of 5.5ghz i run daily 5.2ghz at vcore 1.494 im liquid cooled and temps stay 65 or below C

Originally Posted by sneekypeet

1.494V isnt much better. Keep an eye on what the CPU cores are doing as you run the bench, see if it is throttling Unless this is under "exotic" cooling, that is too much voltage imho for daily running.

And i also have the same problem if i set everyting to stock at 3.5ghz with speed step enabled and memory clocked down to 1333mhz i only loose 3 fps over 5.2ghz with memory at 1866 (stock memory speed) I really think the problem lies in either how the new i7s work with pci-e lanes or something in the motherboards pci-e lanes and the bandwidth getting the signal from pci-e to cpu. atleast thats what evga is saying might be the problem but they have never heard of somone having lower frame rate like this.

[loddie]

the directx 11 benchmark that im basing all my old score vs new ones off of only uses 2-5% of cpu when benchmarking and while running that benchmark my temps do not get above 50c which is why i dont think its cpu related but i could be wrong...

[loddie]

Originally Posted by radrok

But running 16x/4x instead of 8x/8x or 16x/8x wouldn't be a major bottleneck for the second gpu?
That was what I meant, sorry if I explained badly

Clearly this isn't the issue anyway because gpu-z reports 16x/8x

this is the first think evga thought the problem would be but at the same time they didnt see how it would be the problem when there are no 4x slots on the board thats when they had me use gpuz to see what they were running at. they did say i would see a lil loss of fps because in the old set up both cards were runnng 16x and in the new setup one is running 16 and the other is running 8 but it would be enough to notice in game and would only be a little slower in benchmarking

[LordJummy]

I think it would be helpful to know if you are having similar issues with other benchmarks, or just heaven?

I know that heaven also got a lot more demanding over the past couple years as it was upgraded from 2.0 to 2.1 to 2.5 etc. Could it be you are looking at an old heaven score from the original 2.5 or 2.1? Everybody's scores went down after the last couple releases.

Let's see some vantage performance CPU scores, 3dmark 11, etc for comparison. I doubt it's the motherboard at fault if you are not having framerate loss across the board. Even then I'm not 100% sure.

Thanks!