Hello all. Wondering what everyones preferences were for stress-testing their overclock/system stability? I have used both OCCT and Prime 95 while o/c'ing my 2600k, and seem to now have the system as stable as it is going to get.

       One thing I noticed, though, is that they both seem to pick out different instabilities. My Prime runs generally turned up 101 codes after blue-screening; several runs later and vcore tweaks made, I achieved 24-hour stability on Prime95. Subsequent to that, I ran OCCt, and it bluescreened after 30 seconds!!

        It flagged a 124 code, which I researched as QPI/VTT first, then vcore. A few runs later, and the OCCT test passed the one-hour 'auto' test with no errors. Re-ran Prime95 one more time also with no errors. So I am wondering If this is something that you guys/gals do as well? I am really glad I decided to try more than one program. Thoughts?

I like IBT (Intel Burn Test).  I've used LinX in the past (which I thought was a re-packaged IBT but I may be wrong on that).  Also MemTest86+ is great for hammering your memory.

Thanks, Flyinion. I tried IBT as well, and passed sucessfully the first time on 'standard'. I guess it would be safe to assume the system is stable....passes all three stress tests I have tried, and I have run FSX for 5+ hours with Active sky 2, all my saitek gear and panels, in complex aircraft with modest settings and no problems. Coretemp shows FSX heating the cpu to 64C on the hottest core...the rest are a few degrees cooler. Looks like I am in the green!

Use IBT for quick test, if you can stay below 90c, it's probably stable enough. Or, use it to set the limit for your CPU, when you get 90c, that is the limit for your cooler, don't go higher.

If you get 90 with IBT, you will probably stay below 80c with Prime95, and below 60c with FSX(roughly). But this quick IBT test was really usefull for me.  

You have 2600K, so after that Prime95 would be the right choice in my opinion. And for fsx, no matther what you are using for testing, just add a couple of units of Vcore, for full stability.

Sandy bridge - Prime95, Ivy Bridge - AIDA(recomended, but i know people's who used Prime95 with ivy and no issues).

I tried and i dind't liked OCCT at all

I also use LinX, Prime95 and OCCT. OCCT uses a mix of various algorithms so it's a good shotgun test for stability, plus it monitors temperatures and will stop its execution automatically if the configured temp thresholds is exceeded.  BTW, both LinX and IBT are nothing more than a GUI front end for the X86 flavor of the linpack benchmark executable (http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/), so essentially they are identical.

Thanks, Flyinion. I tried IBT as well, and passed sucessfully the first time on 'standard'. I guess it would be safe to assume the system is stable....passes all three stress tests I have tried, and I have run FSX for 5+ hours with Active sky 2, all my saitek gear and panels, in complex aircraft with modest settings and no problems. Coretemp shows FSX heating the cpu to 64C on the hottest core...the rest are a few degrees cooler. Looks like I am in the green!

 

Oops, forgot to mention, run IBT at "Maximum" not standard to get proper stress testing.  The OC guide I was following for my cpu has you run 3 passes at max and if it passes you can be relatively sure you're stable and increase your clock a bit more.  For max testing it then recommended an hour of testing, but when I first started using stuff like IBT and LinX, the "gold standard" was 10 passes you could consider it stable.  It probably takes me about an hour to run 7 passes depending on my clocks and HT on or off.  When I first built my system and was randomly trying to get to 4Ghz or above I would get through like 7 passes and think I was going to make it and then BAM pass 8 or 9 would fail spectacularly with a BSOD.

Oops, forgot to mention, run IBT at "Maximum" not standard to get proper stress testing.  The OC guide I was following for my cpu has you run 3 passes at max and if it passes you can be relatively sure you're stable and increase your clock a bit more.  For max testing it then recommended an hour of testing, but when I first started using stuff like IBT and LinX, the "gold standard" was 10 passes you could consider it stable.  It probably takes me about an hour to run 7 passes depending on my clocks and HT on or off.  When I first built my system and was randomly trying to get to 4Ghz or above I would get through like 7 passes and think I was going to make it and then BAM pass 8 or 9 would fail spectacularly with a BSOD.

 

 

Cool. I will try the test at a higher setting. If you don't hear back from me, it most likely indicates that my computer melted down the entire neighborhood....

Use IBT for quick test, if you can stay below 90c, it's probably stable enough. Or, use it to set the limit for your CPU, when you get 90c, that is the limit for your cooler, don't go higher.

If you get 90 with IBT, you will probably stay below 80c with Prime95, and below 60c with FSX(roughly). But this quick IBT test was really usefull for me.  

You have 2600K, so after that Prime95 would be the right choice in my opinion. And for fsx, no matther what you are using for testing, just add a couple of units of Vcore, for full stability.

Sandy bridge - Prime95, Ivy Bridge - AIDA(recomended, but i know people's who used Prime95 with ivy and no issues).

I tried and i dind't liked OCCT at all

Yes, it's important to note that.

 

Both Intel and Asus do NOT recommend the use of IBT, OCC etc, for testing Ivy Bridge. They are not validated for Ivy Bridge. They do not test all parameters effectively.

 

Personally though, I found when overclocking my 3770k, that the latest version of Prime was okay.

 

The bottom line, is to consider why you are overclocking, for what purpose. And it's NOT to achieve stability with ridiculously torturous stress testing software.

 

On the contrary, it's to see how high you can clock, while running FSX and or games with stability.

 

If your goal is to achieve utter stability with a ridiculously stressful utility, especially one that isn't validated for your CPU,so doesn't test all instruction sets effectively, you may find yourself adopting unnecessarily high voltage, that isn't required for everyday use.

 

So the true test, is if you can run the software you usually run, with stability long term.