View Full Version : How many passes of IntelBurnTest to consider stability adequate?
I'm currently working my way through various timings to try and a system 100% stable, and I think I've got a decent set of BIOS settings that I will be happy with - can successfully run five passes of IntelBurnTest with 6GB selected (of 7GB available)
It's looking promising, but a longer run would be more convincing. I guess any system can crash given enough time, so how many passes is generally enough to say this is good enough, this system is effectively stable?
At the end of the day its up to you, if you are folding or running some such program that will load it up to 100% all of the time, then you might need 20 passes (i believe this is what some people say) if not well does it do what you want it to without crashing?
Some people are quite anal about stability of overclocked systems insisting it must run such and such for X amount of time but unless you are running some distributed computing program that absolutely NEEDS to be run on a stable system then it is entirely up to you whats "stable".
I normally run it on max memory for 5 times and i'm always stable.
I will probably end up doing a longer test of a few hours when I can keep one eye on it. I know I few people run overnight tests, but 20 or so passes sounds like a more sensible compromise to me.
I've got 4x 2GB memory, which was purchased as two packets of matched pairs rather than four matched sticks (which costs a ridiculous amount) and I'm not sure if that decision was a mistake. Something definitely isn't quite right, at close to the max specified RAM speed the motherboard is getting a bit flaky and is sometimes turning off 4GB, but then when I turn the BCLK back down I still only get 4GB on reboot. It possibly needs a slight voltage offset between channel A & channel B, but I don't have a volt meter so would't know what difference to set.
On the plus side I got IntelBurnTest stable by adding a DRAM clock delay to the channel B. 4GB only is perfectly happy with 200 BCLK, but all 8GB needs a lower BCLK of 190. I'm hoping to tweak enough to reach 195 stable and then I will be 100% happy and stop messing with it.
I normally run prime95, if it survives a 12 hour run then it's stable. 24 hour run (if I can be bothered) then it's definitely stable
Toonshorty
24-03-10, 20:39
I just run IBT for a few runs (5) or so, if the overclock is unstable it usually says so in the form of a blue screen while it runs.
After that I just pretend it's stable and use my PC, if it's slightly unstable it will start to lag and bluescreen in a few hours time, usually stating it wants more voltage.
Run IBT a few times (should take 5-10 minutes) and then just use your PC as you usually would since theoretically that is the best stability test, who runs IBT all the time?
IBT stresses the CPU more than any other program. As you cannot emulate this sort of load with any application (only OCCT uses a similar testing algorithm) this is an unrealistic scenario. I would say that 5 passes AFTER the temperature of the CPU stabilizes is enough.
Powered by vBulletin® Version 4.1.12 Copyright © 2012 vBulletin Solutions, Inc. All rights reserved.