• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Prime95, i7 920, 3.8 stability?

A

Aumakua

Limp Gawd
Joined
Dec 14, 2009
Messages
333
okay so I thought I had a winner with my overclock until this morning.

Yesterday I ran Prime95 (64 bit) for 10 1/2 hours in blend mode, no errors!
then I ran it in large FFT mode overnight, woke up this morning and heard windows restarting (windows chime), it ran for 8 hours 22 minutes before crashing.

would you call that stable since it went that long in both tests? or does the fact that it did crash after 8+ hours with large FFT means its not stable?

I am doing this at 1.2875 vcore (which actually reads at 1.248 on cpuz), I tried goign up to 4.0 ghz but was having to raise the vcore way to high for it to make sense (power/performance).

My settings on a X58A-UD3R board are:

200 bclock X 19
1.2875 vcore
1.30 vtt
1.34 IOH (SLI)
1.20 ICH
1.64 RAM (ram is supposed to run at 1.65 but I can't set it at odd numbers)
QPI at 3600 mhz
NB at 3200 mhz (16X, double my memory setting)

Ram is OCZ gold 1600, which has passed all tests, running at spec 8-8-8-24 timing
although one of the settings is on auto and bios shows 60 but others online said to set that to 88, haven't since I wasn't having problems

also performance is set to Turbo in bios (again did not switch back to standard because I wasn't having issues)

no vdroop
ht on
turbo off

temps were 67-65-65-64 with realtemp at max load

I don't want to raise the vcore more if I don't have to or mess with RAM settings, but kind of bummed me out about the large FFT failure after 8+ hours, especially when the blend went over 10 hours without issues. suggestions?
 
Those temps seem high for that vcore, what cpu cooler are you using? Is your airflow sufficient?

EDIT: gonna have to retract my comments about the vcore, I am not sure if that is high for i7s, it may be. Also, I subscribe to the school of thought that if you can crash an overclock with anything, it is not stable :) Try out IntelBurnTest and you may find your "stable" overclock isn't so stable
 
Last edited:
yeah I was going to download LinX and see what that does, usually with prime if its memory it gives errors and doesn't crash, crashing usually is from the cpu *just in my experience*

I have a dark knight cooler, mx2 thermal paste, and 9 fans in my system included an exhaust behind the cpu (spedo advance case), I don't think airflow is a problem, when you put your hand in the case its a pretty cool breeze.

when I was running at 1.25 vcore (1.184 actual) it never passed 60 temp, I am sure if I go 1.35 which I think I would have to for 4.0ghz stable, it would be over 70. Heard not to go over 70-75 to be safe, max 80-90 on the i7 but they do run hot.
 
phreak144 said:
Those temps seem high for that vcore, what cpu cooler are you using? Is your airflow sufficient?

EDIT: gonna have to retract my comments about the vcore, I am not sure if that is high for i7s, it may be. Also, I subscribe to the school of thought that if you can crash an overclock with anything, it is not stable :) Try out IntelBurnTest and you may find your "stable" overclock isn't so stable
Click to expand...
His temps are right where they should be for that vcore and clockspeed.

My guess is that either your vcore or your vtt (or maybe both) needs to be bumped up a notch or three. With your cooling you should be able to bump it up to 1.35 for vcore and 1.35 for vtt. Anything below 75 degrees while running a stress test is golden and I would think you should stay within that realm. If you get to 4.0GHz stable (or maybe try for higher if you like) try dropping the voltage down until you find the lowest point. Use Intel Burntest or another linpack stress test to speed up your testing. Once it passes 10 runs of a linpack stress test turn things over to prime95 over night.
 
thanks for the feedback, was really hoping to leave the vcore and vtt alone as I have seen others reach 3.8 ghz at around those numbers and its really a sweet spot, hardly above standard but able to boost to 3.8 ghz, to get to 4.0 you really need to bump them up, I might play around at 4.0 but will ultimately run it at 3.8 everyday. I will slowly try to up those two after I install LinX, hopefully I can keep vcore 1.30 or less and vtt at 1.34 or less for max performance/power ratio
 
I would say it is stable enough that you would never have a crash while doing regular computer stuff including playing games.
 
Those temps are pretty good, no need to worry. I never feel comfortable running a system that isn't 100% stable. It could crash on your at the absolute least opportune time.

I would bump VTT a notch or two.and retest. A slight bump in CPU PLL volts may help too. Also do you really need that high of IOH? Max recommended Intel spec is 1.20v and I've never really seen anyone run it that high. I don't see why you need more ICH voltage either unless you're running a lot of SATA devices or a huge RAID array or something.
 
Interesting, so I downloaded LinX and ran it for 20 tests (or passes or whatever) at full memory and it passed without a problem (ran for about 1 1/2 hours).

might retest the large FFT with Prime95, but I think its stable if it passed LinX and 10 1/2 hours of Prime blend.

again thanks for your guys help/feedback on this
 
You must log in or register to reply here.
Back
Top