Originally Posted by Maelthras
Running 20 passes of ibt is much more demanding than any game. It's long enough for him to hit peak temp and stress the cpu to beyond a real life scenario. People that run that crap for 24 hours are literally just wasting time and electricity. So if you run for 24 hours, you get a bsod, what caused it, do you know, no. You know nothing about what caused it. As I stated before there is not really a way to tell the difference between a memory error or a unstable overclock. So ultimately the entire process is pointless because it doesn't tell you crap, just that something went wrong.
Running 20 passes is perfect, if someone can pass that then their overclock is stable.
Glad it works for you.
I believe that a blend test for a long time is a better indicator of stability than any other test. Not only my opinion but not necessary the right one. Do whatever you like, but let me suggest what works for me.
I dont fold so a 24h stress test is a minor dent in my pocket. And then i keep c-states on. Definitely electricity waste is the last concern for such short time.
You should do ANY stability test with memory at stock (and possibly you test the memory first thing before doing any overclock), to exclude that.
But I would convene that if it is a newly assembled system it could also be some driver or windows bug. A good idea would be to run the system for a week at least before overclocking it, but the temptation is too strong