Sticky overclocking failure

[rmongiovi]

I've noticed this since 1701. I judge my overclock with 2 hours of OCCT. I start with settings which pass that test, and then decrease voltages until it fails.

What's odd to me is that frequently, when I increase my voltage back to what had previously been 2 hours stable, OCCT continues to fail rather quickly; normally within a few minutes. It rarely lasts a half hour. Further voltage increases do not regain stability.

But if I clear CMOS and re-enter the values I had just before the initial failure they are once again stable for 2 hours. Somehow or another the OC failure acts like it's "sticky" and clear CMOS re-initalizes something that lets a previously working overclock work again.

Has anyone else experienced anything like this?

[meankeys]

Its possible its your memory timings. Each time you reboot the memory goes thru a training seq. if you use auto your setting may change each time you boot. also if you change your block feq. allot. try saving your settings to a profile. them clear the cmos then select that profile once you enter the bios. this should help retain the integrity of your Oc

[Qwinn]

I can say if there *is* such an effect as this, it would explain a lot of weirdnesses I encountered during my overclocking that I was never able to find any other explanation for.

I thought something along those lines might be happening at the time, actually, but the added load of having to clear CMOS and do a full reset between every change in settings was a little too grueling for me to do consistently (especially when trying to find a better System Agent voltage - ouch).

[rmongiovi]

It feels like an uninitialized variable to me. Clear CMOS sets it back to 0 and you start fresh.