Conflicting Stability Reports

[NeonLights]

I'm trying to dial in a good OC on my 4690k, and I'm getting conflicting issues. My starting point was 4.6ghz @1.20v, passed IntelBurnTest Maximum setting and an hour of Aida64 just fine. Bumped it up to 4.7ghz, 1.27v would pass fine on Aida64, but I had to kick the voltage all the way to 1.3v and still couldn't get a pass on IBT standard! I poked around on some forums and installed Realbench, and it passed 15 min. of stress testing on 4.7ghz @ 1.27v. I decided I'd had my fun for the day and dropped it back to my original 4.6ghz @ 1.2v, decided to run IBT on the Very High setting just to be safe, and it failed! I now have to have 1.235v to get a pass on IBT, whereas 1.22v is the new minimum for aida64 and realbench. Any ideas as to what the heck is going on?? Is IBT suddenly much more demanding, or are Aida64 and Realbench not stressing enough? As a side note, I got OCCT and ran it for 10 minutes, 4.6@1.22 seems to work fine.

[Oshino Shinobu]

It's not conflicting, it's just showing that the system is not stable. If it fails any of the tests, you need to adjust voltage or back off on the clock until it never fails. 

 

EDIT: OCs can pass tests once, but fail another time. If it fails, even after 5 times passing, it is still unstable. 

[W-L]

-SNIP-

 

That's pretty normal the system initially wasn't stable with only a hour of testing you need o stress test for long say 12-24 hours if you want almost guaranteed stability.

[NeonLights]

It's not conflicting, it's just showing that the system is not stable. If it fails any of the tests, you need to adjust voltage or back off on the clock until it never fails.

EDIT: OCs can pass tests once, but fail another time. If it fails, even after 5 times passing, it is still unstable.

So should I even bother using the other tests since IBT failed so quickly where others were fine?

[Oshino Shinobu]

So should I even bother using the other tests since IBT failed so quickly where others were fine?

It's a good idea to use a variety of tests. IBT and Aida64 should be enough though.

[NeonLights]

That's pretty normal the system initially wasn't stable with only a hour of testing you need o stress test for long say 12-24 hours if you want almost guaranteed stability.

I was just wondering if there was an issue since I passed IBT on maximum no problem, then later couldn't even pass on standard with the same settings

[Imakuni]

So should I even bother using the other tests since IBT failed so quickly where others were fine?

Yes you should. Maybe not at first, but after you pass your IBT, give it a check with the other tests as well.

[NeonLights]

It's a good idea to use a variety of tests. IBT and Aida64 should be enough though.

Okay thank you for the advice

[W-L]

I was just wondering if there was an issue since I passed IBT on maximum no problem, then later couldn't even pass on standard with the same settings

 

It not that you should disregard the other tests but test for longer say in Adia 64 or Intel XTU, the Intel burn test is a good test to hammer it with sudden loads.

 

It's same with Adia 64 or others where if you test for 12 hours and it doesn't fail the first time but if you go again and it does then that means it wasn't stable even the first time around but just happened to not fail for 12 hours that time around.

[NeonLights]

Yes you should. Maybe not at first, but after you pass your IBT, give it a check with the other tests as well.

What setting would you recommend? Is maximum absolutely necessary or will high/very high produce similar results?

[Imakuni]

What setting would you recommend? Is maximum absolutely necessary or will high/very high produce similar results?

Depends on how hard you want your system to be stable. I'd start with a very high for the basic testing, and once you are happy with your results and ready to just do the final benchmarks and be done with the OC, then you could try max, just for the sake of it.