Originally Posted by TLM-610
Many people here have posted according to their experiences so their opinions are diverse because of various diverse testing conditions. None of them should be discredited.
Try and follow what am about to say anyway.
When you are out to stabilize your OC, the factors to consider are:-
-what do you intend on running your OC for?
-the processor instruction sets in the CPU to be OCed (CPUz will reveal all of them to you).
-the release version of the applications running the OC.
Here is the reason why the above, there isn't a perfect testing application out there that will give you the perfect results that you are looking for. You will notice this by the variety of testing methods people here have adopted in this post in order to really get to that stable OC. You will realize that those who engage in a battery of tests get the best results meaning that they haven't ruled out any testing application, they just understand that one tests better at one thing than the other and if they could run them all and come out error-less at the end of every one of the tests the likelihood of running into a sticky situation is very rare as opposed to the ones who have just testing using one application. Some here have experience, like the ability to run prime for 2 hrs observing keenly how the system responds and quickly swap to another testing application that you can only get if you employ a battery of tests having run them like a newb at first to really study your system (running prime for 12hrs+), them with an intelligent guess give your system certain settings and run them for the 2 hrs and move on. These guys may not point out that to you, but sound overclocking is all about valuable experience.
Each new release of processors per generation contains an update of processing instruction that are to be used by the applications that you engage your processor to run. It is very vital that the responses between the application and the processor are timely without delay or a internal parity processor error will be logged which will progress into a full fledged BSOD. It is wisest to note the release date of your processor first, then note the release date of your OC testing application. The best is to have the release date of your testing application be more recent than that of your chip. It would really be unwise to test a present day processor with a dated testing application (your definitely not going to be stable even if the testing app proves complete stability beyond reasonable doubt which I suspect some here may have experienced and written off a testing app just for that reason). It is also equally important to note the changes that the logs of the testing app show in your present day testing app over previous versions, to see whether there is real support for your testing application and give you a feel of how solid its results might be at the end of the test.
If you intend to run your PC for applications such as sound production, graphic designs and media ripping. Your OC wouldn't need so much juice to term it stable, not like someone who is folding. It won't necessarily need to be pushed to the very ends of stress to have those applications receiving their desired responses from your processor. This kind of stability is application specific and may not count to others as being stable for your system. It is an engineers approach to cutting costs as much as possible doing away with what isn't necessary like a H320 for a mild overclock.
If you intend on gaming, then the above stability won't cut it, you will from time to time BSOD crash. The many flags that have been pointed out such as WHEA errors, hanging, drop in FPS and crashing applications a few but just to mention also do come with experience and studying your system as you go. They mean something when they happen and the smart ones can decipher whether to back down on VTT or up the vcore. If you intend on Folding@Home, you are in for some very extreme measures, simple runs will not cut it. You may have to run pretty lengthy tests to acquire full stability and even resort to to of the line cooling measures and better performance parts to deal with the system's hurdles. There is quite abit more that I haven't mentioned mainly because some here have already mentioned in conjuction with what I have said, I'm just hoping you can piece it up smartly and get the gist.....
So really a stable OC may take quite a while or just a few hours to really be deemed as stable but predominantly depends on your range of applications that you will use with that desired OC.