i2500k and ASUS P8P67 Pro Overclocking Voltage (Vcore) Issues

KingOfComputing

Honorable
Mar 17, 2013
94
0
10,630
I have a i2500k on an ASUS P8P67 Pro that I have owned for about a year and a half. I originally got it as an overclocked set with memory and a Corsair H50 water cooler. It came overclocked at 4.6GHZ though the seller assured me he has had it at 5GHZ no problem at all. Problem is the BIOS somehow lost the OC settings on delivery and the seller hadn't made a profile.

I wasn't too bothered at the time as I thought if the chip is good for 5GHZ, overclocking it should be a piece of cake. However, I have always found this CPU awkward to overclock needing extra voltage (CPU voltage) even to achieve basic overclocks like 4.3-4.5GHZ, which many chips will do happily at stock or 1.3. I have tried following guides to get a good basic 4.5GHZ overclock with reasonable voltages like 1.3 but my CPU always needs more than the guides say. The only way I can get it to do 4.5GHZ is by setting around 1.36 or by using 1.30 and setting the loadline and phase at their maximum "extreme" settings but then in Windows CPU ID and Core Temp actually report the voltage is overshooting the 1.3 target.

4.6 requires 1.39 and 4.7 needs over 1.4 (about 1.43). With these kinds of voltages being needed how would the chip ever reach 5GHZ without burning up? I can run it at 4.7GHZ stable but temp wise it is pushing things as on Prime 95 small fft it reaches 85C after several hours and I don't really want to go higher, not even sure if that is safe for a 24/7 overclock?

Am I doing something wrong? is there anyway of getting more out of the CPU without bumping it's voltage (or at least mitigating the voltage bump)?

I have seen guides suggesting setting a very low CPU voltage initially (1.2) and turning the CPU Current Capability up to 140%, could this help to keep the CPU voltage down?

Also is there any consensus on PLL over voltage setting enabled or disabled?

Finally, is it normal to have big (10C+) variations between core temperatures? Cores 0 and 3 always seem a lot cooler (~10C) than cores 2 and 3. I have confirmed this in both CoreTemp and RealTemp.

Any advice at all much appreciated, system specs follow:
Cooler Master CMII Full Desktop Case
i5 2500K
Corsair H50 CPU Cooler (with Corsair fan on each side of radiator)
ASUS P8P67 Pro
16GB (4x4GB) 1.65v Kingston 1600MHZ
100GB STEC SSD
1.5TB HDD
Pioneer SATA DVD RW
AMD Radeon HD6970 (with custom 3 fan cooler)
ASUS Xonar DS 7.1 Sound Card

 
well on the h50, 85 C is a fairly normal temp at 4.7, though you should not need that kinda voltage for those clocks unless you just plain have a very poor overclocking cpu (typically sandy was great for OC though

i would say completely clear the cmos and all motherboard settings and restart by just increasing core multiplier and only touching the vcore and vdroop offset, also turning turbo boost off
 

KingOfComputing

Honorable
Mar 17, 2013
94
0
10,630


Is the H50 a poor cooler then? I know it's bottom of the range as far as Corsair's water coolers go but I still thought it was pretty decent? Do you think taking it off and replacing the thermal paste could help? Would it (the paste) be likely to have degraded in the year and a half I've had it?

RE: The turbo boost, I can't turn that off as it is how Asus boards (well the P8P67 Pro at least) overclock, the overclock IS effectively the turbo boost while the standard multiplier remains 33.
 

KingOfComputing

Honorable
Mar 17, 2013
94
0
10,630


Wow, that is so different from my experience with the CPU. If I set it to 4.5GHZ and 1.2 voltage I would be surprised if it even completed POST and if it did I'd be looking at a blue screen in seconds, even at 1.3 it needs the maximum LLC. Is there anyway I can tell if my CPU is just a really bad overclocker, is it specific model numbers or revisions that are bad for overclocking?