Originally Posted by 66racer
Honestly stability is relative in some ways. Use the pc a bit and see how it does gaming and such. I usually do a few different prime test for a total of 4hrs and have never had issues with a stable overclock. Im actually downloading that Asus program/suite now to see how it works.
Basically if it doesnt crash on you....technically its stable
It depends on your work load, and gaming usually isnt too hard on a cpu in comparison to a stress test.
During my time here, I've seen things that force me to disagree with this. For example: "My rig has been stable for 6 months and now I'm seeing these problems. What the hell is going on? Is my motherboard or CPU dying?" I usually ask them what they've done for stability testing, and the answer I always get is something to the effect of "Yeah, it is perfectly stable. I tested it for 6 hours in Prime95 and it was fine the entire time. So, what's going on? Do I have to RMA something?" My answer is always, "No. You just didn't test the stability of your system well enough."
With subtle instability like that, not only can you run into problems later on (a day later, a week later, a month later, a year later, etc.), but you can have data corruption. For this reason, I will always recommend the most demanding and extreme stability testing - especially with a respectable overclock. For something tiny like 4 GHz for Sandy/Ivy/Haswell,, pff... I would almost say don't test it at all and just let your normal use be the test. As you overclock higher and higher though, it becomes more and more important to test the stability properly and thoroughly.
I remember overclocking my E2180 to 3.0 GHz. Everyone recommended 9 hours of Prime95's Small FFTs test, so that's what I did. It stayed stable for all 9 hours, so I thought that I was done. However, I was still seeing minor problems that looked like minor bugs in some of my programs. So, I decided to see if my system would stay stable for 12 hours. Well, lol, it failed just after 10 hours. So, I worked towards making it stable enough for 12 hours. I ended up making past 12 hours and I even let it go for 14 hours and then I stopped it because I wanted to use my computer. The result? Those minor bug-like problems completely disappeared.
Today, I understand why: there are dozens and dozens of FFTs that Prime95 can go through, and each one is different from the rest. In Prime95 27.9, the Blend test has 82 FFT sizes and it takes about 24 hours to get through all of them at least one time each. So, with Prime95, the length of time isn't as important as most of us think.
Edited by TwoCables - 5/14/14 at 4:50pm