Ok, I have some updates.
First, I haven't been able to find a stable setting for 4,6GHz for my i7 2600k with 1,4320V even though I had it stable on BIOS 1815 with only 1,3750V static setting and high Vdroop (after passing 24 hours of Prime95 small fft).
Another thing is that VDroop is always present on any setting. Now I have set it to Low (Performance) and I get a Core Voltage of 1,4000V at load from 1,4320V in idle.
With version 1815 the VDroop setting was reversed: high (power saving) was generating a very steady voltage under load (very few times it dropped by 0,005V) and low (performance) was generating a high drop in voltage under heavy load (about 0,030V).
Now I'm at 4,5GHz with 1,4320 in Idle and 1,4000 at full load and Vdroop setting to low (performance).
I regret updating the BIOS to v1900.
You clearly haven't read Performance Tuning guide and other guides for this board and CPU overclocking. Unless you go over 80C there are no concerns about cpu life and there are a lot of reports that this CPU can sustain 1,55V 24/7 with proper cooling.
At the configuration mentioned above - Idle Vcc 1,4320V and load Vcc 1,4000V - my temps are 35C in idle and 70C Prime95 small fft and 50 when gaming (Thermalright Venomous X cooler with Scythe Gentle Typhoon fan).
So I'm not worried, just dissappointed with this bios because instead of improving OC capabilities it practically crippled it.
Just an update on the current BIOS (v1900) and the replacement board I received (with the P67 PCH bug fixed).
The VDroop on this board is higher. Set on low vdroop, and fixed voltage of 1.4450 -> without load i get 1.4520V and under load i get 1.4000V
If i set the vdroop to high then i get 1.4480V on idle and 1.3650V under load (that is nearly 0.1V Vdroop).
It is clear to me that intel boards aren't meant to compete with overclocking boards from other manufacturers, although with bios v1815 i had better results (higher stable CPU frequency with lower voltage and with nearly 0 Vdroop).
I really hope they release I better BIOS in the near future.
I am having a similar problem too. Today I decided to do a modest overclock on my 2500K. I got it up to 4.5GHz with all of the turbo multipliers set to 45 and the Processor Core voltage set to 1.4000. CPU-Z shows the correct voltage of 1.400 when the PC isn't being stress tested. Once I run Prime95 or Intel Burn Test, the Core Voltage drops to 1.368V. This is with V-Droop control set to Low V-Droop (Performance). Also CPU-Z shows the multiplier as x43 even though I have it set to x45 for all cores in the BIOS. Why isn't it running at x45 when I run a stress test?? Do I need to be worried about the drop in voltage when stress testing? I am a bit new to overclocking but I really want to learn. Also I hope soem Intel techs are reading our posts and letting the engineers know what we are experiencing so that they can work on BIOS updates that will allow this board to be an amazing overclocker.
Is there some way that we can contact the techs??
The previous post is from me...sorry but I didn't realize I wasn't logged in when I posted that message:
Anyway even though I get some big VDroop when I stress test, my overclock has been stable and I haven't crashed or BSOD or anything else. What's the best way to test my OC for long-term stability? Prime95 or Intel Burn test?
High voltage droop pushes you to increase the Vcc. So when idling you have a higher voltage than when you are loading the cores (oposite to what you need). I had very good results with bios v1815 with High Vdroop setting (in that version, the setting was reversed), but not in v1900.
I can think of two reasons your multipliers don't reach your setting of 45:
1. Maximum TDP or TDC value is too small, try to increase those in BIOS
2. CPU is overheating passing the 85C mark
I mostly use prime95 for stability testing.
Thanks for the advice. I changed the TDP to 180 and the TDC to 200. I also increased base clock to 105 and I kept my ratios at 45. I was able to get to 4.73GHz at 1.4V (1.368 with droop) but it's stable so far. I have to do further long-term stress testing with Prime95.
Temps are good, I think, and I wonder if I'll be able to get to 5GHz?