Originally Posted by Wishmaker
So, you are running 1600 MHz ram on that thing. Your RAM clocks are 808.0 x 2 = 1616 MHz. You then decided to manually play with your UNCORE and raise it to a point where it sits at 7679.8 MT/s. This is roughly 3837.9 MHz as shown in the above screenshot.
The rule of thumb for stability is 2:1 which leads to 1616 x 2 =3232MHz. Now if we convert this int MT/s , 6464 which is more than stable on pretty much any X58 system out there with any chip. Having your chip running like this will lead to a decrease in performance far greater than the increase in BCLK. Nobody runs it at the minimum frequency and we all aim to raise it higher. I am running it between 3500 and 3700+ depending on how high my overclock is.
You are running 1600 MHz with an UNCORE meant for 2000 MHz RAM. Stability taken in consideration of course. Can you please run IBT with memory on HIGH or Extreme to see how it holds together? If it passes the first run I would be surprised. My chip to pass IBT at 3700 needs 1.37 QPI and 1.67 on RAM! Which is above the safe voltage. I was IBT-ing at 4650 MHz at one point and it crashed after the 8th run! My RAM is rated at 1600 MHz
As long as you do not exceed 1.35V on the QPI, you can run it at over 9000, however, having it stable, that is another matter. You have unlinked certain parts in your overclocking chain which have well defined relationships.
Over clocked to 4.65 GHZ, I'm running most of the voltages on AUTO so was a bit concerned about QPI/Vtt so manually set it to 1.335Volts. So I did 10 runs of IBT at HIGH setting as it won't run on Very high as I am using 6Gb RAM.
Temps were holding at about 75 deg with an occasional peak of 80 Deg.
I did another IBT run at 4.65Ghz with QPI/VTT 1.355 Volts at NB/UNCORE 3837 MHz and RAM at 1616 Mhz CL 8 8 8 24 88 triple channel. Best Cinebench CPU of 1089.
Seems stable. What do you reckon?