I come from an older school of thought when overclocking, and that is change one thing at a time and ease into it. Jumping straight for a high clock with high voltage is going to be a problem. Proof of this would be that you're maxed on vcore and only at 4ghz. I'm working on a log right now in the x99 forum and it seams extremely basic and slow to produce result's, but you might want to check it out. This morning I hit 4ghz on my 5820k @1.10 vcore. I stress tested last night at 3.9ghz and previously 3.8 3.7 3.6 all at 1.10 vcore. My method here is to set a certain vcore a bit over stock and slowly bump the ratio into it, when unstable add a little voltage. It's how I did it when I started and how I prefer to do it, I don't use adaptive or speedstep or turbo boost or any of that. There are methods that work using those but I prefer not to.

Basically your method is lacking if you got that voltage at that speed.

@JustinThyme What a answer for a first time poster 😞 , but ok, i will try to explain some things. First, i got a very good deal for the 5960X and the Strix board, second my intention is not overclocking to win a competition or to show how big my penis is. I want to build a workstation and have a effective machine for my adobe suite and cinema 4d. I want to be able to overclock the machine with a single click, if it comes to rendering the work, which is the most CPU and timeconsuming process in my work and i think this is a understandable reason. I read many reviews and no one mentioned that a 4Ghz overclock with a 5960X needs a non-plus-ultra mainboard, that alone costs 500-600€. So i excuse my naive thoughts, that a 4Ghz overclock that is even a whopping 500Mhz under the "average" expectation, is doable with my crappy hardware.

Regarding the watercooling, this is what is written there: "Depending on your ambient temperatures, full-load voltages over 1.25Vcore fall into water-cooling territory (dual-radiator). With triple radiator water-cooling solutions, using up to 1.35Vcore is possible. For air cooling the value is much lower, limiting total overclock, so plan your cooling investment appropriately."

Well the 115i is a dual-radiator and a huge one too. Nowhere it is stated that you need a custom-build water cooling set, maybe someone should add some line here, that AIO sets are only garbage and commonly known for that.

Regarding your leaderboard, well look at it: http://rog.asus.com/rog-pro/realbench-v2-leaderboard/?ipage=1&Laptop=desktop&CPU=intel&VGA=NVIDIA&cp...
And tell me what you see at the 11th place there, on my side of the screen, i see a "STRIX X99 GAMING Intel Core i7 6950X 4199.0 MHz 1599.7 MHz". So that is a 10 core CPU which i guess is even harder to overclock, at a 4200Mhz overclock with RAM speed settings same as mine. Sadly i dont see what kind of cooling-magic is used to achieve that, but i still hope you maybe understand a little, why i have assumed i could achieve the same with a older 8-core CPU.

So to summarize it, the linked guide gives me the conclusion, that it is a big misleading horse**** and nothing else. 😮
I still thank you for your deep insight and for telling me what the VRM is, but not where it is placed. I guess it is over the cpu socket and the reason why i cant have stable 4000Mhz, as reaching nearly 60 degree celsius for VRM seems a little to much. My cpu temperature is between 72-78 celsius under heavy load. I dont get blue screens, i just cant pass the stress test (1hour). It just stops after about 15-20min.

I use ROG Realbench for the tests. I have the XMP switch enabled on the board and in bios too. I must admit that the switch alone will not work, to enble the XMP speed and timings. Didnt touch the jumper, cause i didnt consider 4.000Mhz as extreme overclocking. As you all say that 1.35 core voltage is high, i will try to lower it to 1.25 and see what will happen 🙂 . 3.900Mhz is stable for sure, didnt had any BSOD or hangs etc. while stress testing or working. Problems start with 4.0Ghz and with 4.2-4.4 it will become really unstable with occasional blue screen and so on.

Kiffalot wrote:
I use ROG Realbench for the tests. I have the XMP switch enabled on the board and in bios too. I must admit that the switch alone will not work, to enble the XMP speed and timings. Didnt touch the jumper, cause i didnt consider 4.000Mhz as extreme overclocking. As you all say that 1.35 core voltage is high, i will try to lower it to 1.25 and see what will happen 🙂 . 3.900Mhz is stable for sure, didnt had any BSOD or hangs etc. while stress testing or working. Problems start with 4.0Ghz and with 4.2-4.4 it will become really unstable with occasional blue screen and so on.

See if you can get your CPU overclock stable without enabling XMP. In other words, run your memory at stock defaults (2133MHz).