Quote Originally Posted by masterkaj View Post
I ran prime small fft for 8+hours, blend for 6+hours, OCCT CPU for 4+hours, Ram for 6+hours... OCCT Mixed fails around 2hours for some reason. I tried upping my CPUPLL to 1.6v but it failed after 15min.. when I go back home monday I'll try to lower it a bit and see how it does.

Don't know what setting to touch though.. why would it fail on the mix but pass the other two? I don't think heat is an issue considering it (NB/CPU) does not get any hotter than when I run each test separately.

I was just wondering why do the tests fail when try to up my NB or FSBvt? Temps are not out of the norm so why is it so sensitive. I figured I could up my FSB to 1.5, NB to 1.65 and bring it down from there... but that fails pretty quick too. This setup I have currently have 1.3fsbvt(why does anandtech say minimum of 1.5?? for 440+fsb) and 1.59nb is the most stable this OC has been so far, but fails that damn mixed test after 2 hours.
I think you have to run OCCT in RAM mode again for a longer time, and a Prime95 Small FFT for a 8h also. I'd go with a long small FFT first to rule out vcore/vPLL/vFSB issues
Then, if stable, run a new OCCT RAM for at least 8h. If unstable, it is something to do with RAM timings/RAM voltage/vNB/NB-Strap/vFSB...

But yes, hard to say what it is exactly unless you debug it. OC needs now much more parameters unless intel improves its NB clocks and building (65nm instead of the 90nm) as it is getting a very limiting factor