X48 Rampage Formula Preview.

[Alien Grey]

Thanks, at the moment im on 902 and i did fidle around with the skews.
Now i got:

CPU Clk Skew = Delay 200PS
NB Clk Skew = Delay 100PS

Cant really get more stability with it tbh, probably have to play with it more.
Just failed OCCT after 3 hours on 3655 Mhz , bumping up my NB Voltage.
Ill keep you updated.

Hmm... Does OCCT saves a log somewhere?

Without a log you can't know what voltage to change because you don't know if it's the CPU or DRAM/NB that's causing the instability.

[leo27]

i just saw your everest settings and to me your FSB term looks very low, i need mine at around 1.3 to be stable at 425Mhz Fsb

[Earzz]

Hmm... Does OCCT saves a log somewhere?

Yes everest shows up pictures after completion/failure but atm it doesnt show weird voltage drops, only when stress testing starts but those voltage lines will remain stable.

i just saw your everest settings and to me your FSB term looks very low, i need mine at around 1.3 to be stable at 425Mhz Fsb

Yep you noticed very well, atm its at the lowest setting in the bios, FSB Termination Voltage [1.20].
Untill now it didnt need any more voltage, ill try to run OCCT one more time @ 3655 Mhz because raising my NB voltage last time didnt help and it failed after 2 hours, now i raised the VCore on step.
If this doesnt help i start raising the VTT.
PS: i just saw you are on a 65nm CPU and im on a 45nm, that could explain that you need more VTT

[Alien Grey]

Yes everest shows up pictures after completion/failure but atm it doesnt show weird voltage drops, only when stress testing starts but those voltage lines will remain stable.



Yep you noticed very well, atm its at the lowest setting in the bios, FSB Termination Voltage [1.20].
Untill now it didnt need any more voltage, ill try to run OCCT one more time @ 3655 Mhz because raising my NB voltage last time didnt help and it failed after 2 hours, now i raised the VCore on step.
If this doesnt help i start raising the VTT.
PS: i just saw you are on a 65nm CPU and im on a 45nm, that could explain that you need more VTT

I didn't mean a log for the voltages. Is there a way to know where it failed like in Prime95 Large or Small FFT's?

Try with FSB Termination Voltage at 1.14V or 1.16V. Try to get it LinX or IBT stable for at least 25 passes before you try it again with OCCT or Prime95.

[Earzz]

I didn't mean a log for the voltages. Is there a way to know where it failed like in Prime95 Large or Small FFT's?

Try with FSB Termination Voltage at 1.14V or 1.16V. Try to get it LinX or IBT stable for at least 25 passes before you try it again with OCCT or Prime95.

OCCT just fails and ill get a windows error or it fails and it says which core failed.
I honestly never played to much with Prime95 so i dont know anything about those logs, what i can remember from Prime is that it just stops when a core fails.

I got my system now OCCT stable for 8 hours.



tbh i dont want to run Linx or IBT since from now on my voltages are at that level that the temps are going over the 71C with using them.
So from now on OCCT only.
Im gonna run a FSB of 435 overnight so hopefully ill be back on my old speed by tomorrow, and hopefully higher later this week

[leo27]

If i were you i would use Prime 95 as well as OCCT if you don't won't to use a linpack based test to double check your stability. I've found P95 more reliable than linpack anyway and it doesn't cause insane temperatures.

It really is very simple to use:
If you run a largefft test and it fails it will be due to NB or DRAM voltage.
If you run a smallfft test a failure is normally due to CPU voltage or the GTL's. And if you run P95 Blend+Unreal Tournament 2004 you'll put even more stress on your system (especially NB). If this fails simply check if it fails in large or small fft's and increase the relavent voltage.