soulesschild
Supreme [H]ardness
- Joined
- Feb 18, 2007
- Messages
- 6,176
OCCT is no longer what I'd consider reliable. My original 1.296vcore for 4Ghz (fairly decent for C0's) passed 2hour OCCT easily (which means it passes Prime even easier).
Enter L4D - after 4hours continuous play, I'm treated to the first BSOD. Mind you I had been playing TF2 and CS:S for hours at a time since Feb 2008 on what I considered a stable overclock. After first I thought it was my RAM. Nope. Changed divider, discovered (lol) that this it does 500Mhz easily, so not the RAM getting flakey. Next was to test vcore. Bumped it up two steps.
Done I thought. Nope. Treated to a Windows file BSOD after 2 hours of L4D. Again, this is showing that CPU-based stress tools do nothing to prove 3D stable. Mind you, with the vcore bump I tested up to 2hours OCCT stable to 458FSB (4.122Ghz). L4D was failing at only 4Ghz!
Finally I figured the only thing left to bump was NB voltage. VTT, both in my past experience and notes (and retested again) does not help stability for the e8400's, particularly at such low clocks. Bumped my NB voltage 2 steps (1.37v to 1.42v). Haven't had a BSOD since. So yeah, OCCT 2 hours up to 9x458, but BSODing in L4D at 9x445 (4Ghz)?
CPU tools tell you jack about real world stability. 3D stable is the ONLY thing that matters if you actually want to use the damn thing. Forget about the e-peen, suicide shots of SuperPI. Ask them if it's 3D stable - I guarantee you 99% of those SS are bullshit for real-world use.
You do realize 2 hrs of OCCT is not enough right? A fully stable OC will pass the following:
8-12 hrs of OCCT
24hrs of Prime
25-50 passes of IBT