The answer to your first question lies in the second sticky post.
I scanned through that sticky to see if there was something I'd missed or if any new developments since late 2011.
IDontCare offers more refined insights to different stability tests, even though my practice was fairly parallel.
I'd say OCCT:CPU is probably the best first quick test of an overclock setting if the memory itself has been tested -- advice seems to be 3.5 hours. IDC points to a regimen similar to mine: Affinitized Linx, IBT, PRime95 (both sFFT and lFFT tests.)
I don't run HCI Memtest instances in Windows. I set aside a week's time on a new computer to run the bootable HCI Memtest CD for anywhere from 500% to 1000% coverage, and obviously, the more RAM, the longer it takes.
Here's my question, which I could not find in the sticky (through the years until you get to the end of it).
I've recently downloaded and installed Intel Extreme Tuning Utility.
The program has to be useful toward "some stress-testing purpose."
I'm wondering if it will find instability quicker than OCCT:CPU. Usually, new programs or developments with the old ones are posted in revisions to the initial sticky post, but it hasn't been updated since late 2011. And all but the last page of the sticky responses span the period through end 2013.
The biggest issue I've seen over the last year focuses on the Devils Canyon or Haswell Refresh processors, the heat and the AVX2 instruction set. So we're looking for a test regimen that doesn't need to press the temperature to the highest level while finding any of some few shortcomings in settings that will cause instability.
Obviously, TechyGeek's Silicon Lottery purchase and the possibility of choosing that path would make this less of an issue.
JUST A FOOTNOTE: CoreTemp, RealTemp -- what have you. These are fine -- always the old standbys, but need to be updated when new processors are released to cover their Tj specs. Some stress programs like OCCT and IXTU have temperature-monitoring built-in, and you can test their accuracy against the old standbys. I just say, if you can get your monitoring and stress-testing in a single program, it is convenient. Further, I've never had a problem with CPUID's HWMonitor for getting the "whole enchilada" to include accurate temperature monitoring. There is a collective wisdom in these forums that running more than one monitoring program at a time is asking for trouble, and may lead to false conclusions of instability in a stress-test.
For instance, some months ago I was stress-testing my i7-2700K system, and forgot that I'd left AiSuite (with its monitoring) to run while I started HWMonitor. A BSOD and reset occurred. I made sure to close AiSuite after reboot to make another pass. The stress-test ran 5 hours without missing a lick.
Moral -- don't jump to conclusions too fast if you had made the mistake of "monitoring-program abuse."
Facebook
Twitter