ASUS P8P67 Pro/Evo/Deluxe OC discussion thread

[SimpleTECH]

Also anyone know what the PCI simple communications controller is?

You need the latest Intel Management Engine Driver.

[DesertFox]

Well there's your problem!

Sandy Bridge isn't made to run at stock speeds.

Ye that's true

[Prestone]

(...)

I have desactivated all the energy savings but the core ratio is still moving between 16x100 on idle and 33x100 on load... Why?

Up!

No Ideas?

[coolhandluke41]

hey ..i'm getting some sucky gflops score,what gives ???

[orion23]

Up!

No Ideas?

Bios 1253 is pretty good at keeping a constant speed, even with speedstep enabled.

Under Bios 1053 or 1204 you need to disable:
Speedstep
C1E
C3
C6

[Prestone]

Thanks Orion23. I think I'm still on 1204, I'll check it at home. I hope a new bios will fix that.

[babalouj]

Can anyone please tell me what the hell is going on with my boards (P8P67 Deluxe) bus speed ? I have flashed both 1204 and 1253 and it keeps happening. 70mhz each time. I have used several different versions of cpu-z and aida to confirm this. My video card keeps getting a driver stopped responding error so I know that its actually running at 70mhz. I have tried changing the bus speed to 101 and 99 and it still boots to 70.

[SimpleTECH]

Can anyone please tell me what the hell is going on with my boards (P8P67 Deluxe) bus speed ? I have flashed both 1204 and 1253 and it keeps happening. 70mhz each time. I have used several different versions of cpu-z and aida to confirm this. My video card keeps getting a driver stopped responding error so I know that its actually running at 70mhz. I have tried changing the bus speed to 101 and 99 and it still boots to 70.

Which CPU-Z versions have you tried? 1.56.2 and 1.56.3 should fix the issue.

AIDA64, even the latest version, still shows the wrong CPU speed for me.

*Edit*

What does Real Temp report?

[babalouj]

Which CPU-Z versions have you tried? 1.56.2 and 1.56.3 should fix the issue.

AIDA64, even the latest version, still shows the wrong CPU speed for me.

*Edit*

What does Real Temp report?

Thank you! I reset cmos and downloaded 1.56.3 and all is well again.

[Prestone]

Ok I get it: when turbo mode is on, the cpu will auto decrease to 16x and when it is off, it will stay to the asked ratio.

Yé!

[Sam_oslo]

Some low-volt action with batch# L040B165 .
Just a little 30-min Prime-run to check the vCore scaling. Besides, it`s good to have a low-volt option for a compact build. I know everybody is pushing the other way, LOL,

2600k @4GHz(40x100) 1.14v 30-min prime small FFT

[crossg]

You need the latest Intel Management Engine Driver.

Thanks.
I had to reload windows due to a faulty HDD and used the old mobo drivers I had downloaded to a usb stick.

@ Sam. A month ago 4 Ghz with 1.114v would be impressive with the older Cpu's (actually still is). I am actually trying to get 4.7 with under 1.33 v and having a tough time. 4.6 to 4.7 GHz requires a big jump in vcore on my cpu.

[Sam_oslo]

Thanks.
I had to reload windows due to a faulty HDD and used the old mobo drivers I had downloaded to a usb stick.

@ Sam. A month ago 4 Ghz with 1.114v would be impressive with the older Cpu's (actually still is). I am actually trying to get 4.7 with under 1.33 v and having a tough time. 4.6 to 4.7 GHz requires a big jump in vcore on my cpu.

Yeah, you are right
I've actually nailed a good sweet-spot @4,5GHz 1.27v for 24/7 while I'm following laboratory-rabbits, LOL.
I will probably push it more when we know more about these CPUs, and how resilient they are with high-vCore over time.

That bump/jump seams to be around 4.5-4.7GHz or most CPUs, but it may vary depending on the batch#.

[Prestone]

Is the 100mhz stock frequency is changeable for the 2500k on a p8p67 deluxe? Can we put it for exemple at 125*40 instead of 100*50 to get more ram speed ?

[deaffob]

No, SBs can't do that much bclk change. However, you can get all the ram speed in the world with bclk 100.

[Prestone]

i know than we can get all the "officials" ram speed as 1600, 1800, 1866, 2000, 2133... but can we set a 1600mhz ram to 1678 or 1721mhz for exemple?

[Sam_oslo]

More low-volt action, and some interesting vCore-scaling observations

4.2GHz(42x100) 1.17v 30-min Prime small FFT


Some interesting observation on vCore-scaling:
4.0GHz = 1.144v
4.2GHz = 1.176v
This means it needs 0.032v (1.176v - 1.144v) to scale for these 200Mhz, or 0.016v for each 100MHz.
Then, 4.5Ghz should need 1.176 + 0.048 (3 x 0.016), also it should be:
4.5Ghz = 1.224v
For 3 days ago, I testes with 1-hour prime and got
4.5GHz = 1.272v
This means, it is not scaling propositionally from 4.2GHz to 4.5GHz

There is a big gap between calculated propositional scaling (1.224v) and the actual (1.272v) vCore for 4.5Ghz. We have to cont on some error-margins too, but this gap is too big for that.

Either my CPU is turned to a Persian carpet (LoL, they get better when you use them) in last 3 days, or there is a vCore bump/jump between 4.2GHz and 4.5GHz.
I will try to test 4.4GHz tomorrow hopefully, and then 4.5GHz again, and it should answer most of it.

[Forceman]

i know than we can get all the "officials" ram speed as 1600, 1800, 1866, 2000, 2133... but can we set a 1600mhz ram to 1678 or 1721mhz for exemple?

You can run BCLK up to about 105 or so, but beyond that it stats to get dicey. So 1600 running at a BCLK of 105 would get you 1680 on the RAM, but I don't know that it is worth the hassle/risk.

[Prestone]

Ok, thanks forceman, i think I will stay at stock bclk of 100... It should not worth all the trouble....

[MaStA]

Yeah, you are right
I've actually nailed a good sweet-spot @4,5GHz 1.27v for 24/7 while I'm following laboratory-rabbits, LOL.
I will probably push it more when we know more about these CPUs, and how resilient they are with high-vCore over time.

That bump/jump seams to be around 4.5-4.7GHz or most CPUs, but it may vary depending on the batch#.

Nailed a sweet spot on your 2006K?

I think you made a typo in your signature.

[Sam_oslo]

Nailed a sweet spot on your 2006K?

I think you made a typo in your signature.

LOL, yeah, now i see that. I'll fix thet now.
I have been repeating the same typos several time now. I don't why, but somehow 2006k falls much easier on my keyboard.

[Sam_oslo]

More low-volt action, and some interesting vCore-scaling observations

4.2GHz(42x100) 1.17v 30-min Prime small FFT
http://img.techpowerup.org/110130/Capture025.jpg[/IMG]

Some interesting observation on vCore-scaling:
4.0GHz = 1.144v
4.2GHz = 1.176v
This means it needs 0.032v (1.176v - 1.144v) to scale for these 200Mhz, or 0.016v for each 100MHz.
Then, 4.5Ghz should need 1.176 + 0.048 (3 x 0.016), also it should be:
4.5Ghz = 1.224v
For 3 days ago, I testes with 1-hour prime and got
4.5GHz = 1.272v
This means, it is not scaling propositionally from 4.2GHz to 4.5GHz

There is a big gap between calculated propositional scaling (1.224v) and the actual (1.272v) vCore for 4.5Ghz. We have to cont on some error-margins too, but this gap is too big for that.

Either my CPU is turned to a Persian carpet (LoL, they get better when you use them) in last 3 days, or there is a vCore bump/jump between 4.2GHz and 4.5GHz.
I will try to test 4.4GHz tomorrow hopefully, and then 4.5GHz again, and it should answer most of it.

Some update on this.
I've tested more, and @4.3GHz 1.192v (1.176v + 0.016v) gives instant BSOD at the start of Prime. The next vCore-notch will BSOD too.
There is a little vCore bump/jump @4.3GHz.
That's not a big one, but there is one for those interested in low-volt setups.

I'm still testing @4.2GHz 1.17v and if proves to real stable, then it may be a better option than @4.5GHz 1.27v for my compact setup.

[st0ned]

I'm a bit stuck in my overclock, if someone have any interesting thoughts I would be glad. So


5.0Ghz with mems @ 933 10-11-10-28 1.65v vmem it's stable ( Linx 25 passes all )

5.1Ghz with mems @ 800 9-9-9-24 1.5v vmem it's stable ( Linx 25 passes all )

5.1Ghz with mems @ 933 10-11-10-28 1.65v get Linx error after the 1st or 2nd round..

more vmem doens't imporve it, either does Vccsa ( 0.925v ~ 1 v ) and Vccio ( 1v ~ 1.2v )


5.1 it's stable with 1.424v, I gave him a bit more and the error still shows, so any advices ?

I really want 933 mem, it takes me from 63ish Gflops to 70ish Gflops...

[Sam_oslo]

I'm a bit stuck in my overclock, if someone have any interesting thoughts I would be glad. So


5.0Ghz with mems @ 933 10-11-10-28 1.65v vmem it's stable ( Linx 25 passes all )

5.1Ghz with mems @ 800 9-9-9-24 1.5v vmem it's stable ( Linx 25 passes all )

5.1Ghz with mems @ 933 10-11-10-28 1.65v get Linx error after the 1st or 2nd round..

more vmem doens't imporve it, either does Vccsa ( 0.925v ~ 1 v ) and Vccio ( 1v ~ 1.2v )


5.1 it's stable with 1.424v, I gave him a bit more and the error still shows, so any advices ?

I really want 933 mem, it takes me from 63ish Gflops to 70ish Gflops...

If @ 933 10-11-10-28 1.65v is stable @5Ghz, then it should do the same at @5.1Ghz, unless your IMC or System Agent is bugging.
On SB the Uncore runs at the same speed as Core, so going up to @5.1GHz will have an impact on IMC and System Agent.

[st0ned]

I'm almost there, I just need some dahm tweek for 1T el dorado, or a better bios