Originally Posted by Paul17041993
that's pretty impressive, is it fully stable and not going above ~65C? I would think the NB needs a bit more voltage for 250FSB, but that's a Z board so I'm not particularly sure how different they behave...
you could try the turbo multiplier, it would drop your all core speeds down but you might be able to get 5.2 or 5.3GHz with only 2 cores used, might need a bit more voltage for that though (watch your temps)
the turbo multiplier has to be 1.5 higher then the default multiplier, or you get some funky clock changes, so set your default to 19.5 and turbo to 21 and see how stable it is, try changing your regulation and voltages if it isn't,
I actually have a slight confession to make, the numbers I wrote down, they were wrong, or at least they were from another profile's overclock, bot my 5.142 GHz overclock. I believe it was gr8sho who stated that the 5.142 GHz would require at least 1.5v to the core, and for stability purposes with IBT, gr8sho was right. 1.48v on that overclock booted into Windows fine and performed "normal" tasks just fine, but when it came to actually stress testing it was much, much too low of a voltage for stability. I lowered the voltage to 1.48v because I am just a little on the fence about running at 1.5v+, regardless of good temps.
The numbers I have written down (the CORRECT ones) for the 5.142 GHz stable test with my old IBT list 1.54v to the core. The max temp I have written down during the burn was 62C. I wrote this stuff down (this was in June sometime...) due to IBT crashing (or blue screens, whichever came first) before I reached the voltage where it was stable (just so I wouldn't forget them on reboot and I knew that they were still in "safe" range for voltage "upping").
As far as the NB voltage, yes, I am a bit "stuck" with this as well. Part of me wants to actually lower the FSB clock down to, let's say, 225 or so and up the multiplier higher to achieve a similar overclock. I mean 250 FSB is an impressive feat as far as I am concerned, but if I have my information correct this is drawing a lot more power from the NB and IMC. Though my NB temps are pretty damned good at this point even under stress test, I don't want to have to go raising voltages just because I have "thermal" headroom. In other words, why juice it up higher if I don't have to, especially considering I may be able to alleviate some of the power consumption from the CPU and NB by simply lowering the FSB and raising the multiplier accordingly.
Thank you all though, this information has been extremely helpful and I do appreciate the feedback!
EDIT: I figured I would string this in an edit rather than make a third new post in a row. I played around with my clock speeds last night and again this morning. I lowered my FSB to 215 and upped my multiplier to 23.5, effectively giving me a ~5066 MHz clock speed. I had to raise the voltage to 1.52v to get this stable. All along I have been using Ultra High LLC settings, if I hadn't mentioned that before. Extreme seemed to cause too many issues, which to me is odd unless I have the wrong picture about LLC....anyway, here are some screens of the stability test which ran with flying colors.
This particular test gave me results (Speed/GFLops) in the 10-10.6 range with the lowest GFlops being 84.5. One thing to note, sometimes I run the test and I get all mid to high 9 second range times with GFlops in the low 90's (92-94), this is at the same clock, nothing changed, sometimes running the test one after another just for giggles. Is this fluctuation normal? I can see a variance of .2 - .5 seconds being normal, but shaving off a whole second per pass?
Anyway, here are the screens, the only difference, again, is that one is with HWiNFO64 scrolled all the way up, the second all the way down to show all temps/voltages. I am also going to load maximum size thumbs here per request of gr8sho. Edited by ThisMaySting - 7/27/13 at 10:55am