i5-2500k Overclocking issues on Asus P8Z68-V/Gen3

i5-2500k
Asus P8Z68-V/Gen3
G.Skill Ripjaw DDR3 1600 1.5v 8GB
Seasonic x660 PSU
Cooler Master 212+

First off, this is my first time ever trying to overclock anything, so please be patient with terminology / instructions on how to get my overclock working.

Followed Theocnoob's guide but not having much luck. Figured I would make my own thread to get more responses from people.

I ran prime95 blend for 9 hours. After about an hour, one of the workers (#4) failed due to rounding error. The other three workers were fine.

I ran Small FFT and after about an hour, one of the workers (#4) failed due to rounding error. The other three workers were fine.

Here are some screenshots of my BIOS settings and Prime95/Hardware Monitor.








Currently I have CPU C1E disabled and Intel Adaptive Thermal Monitor disabled (they are shown as enabled in the pictures, I have ran prime95 with both options).

Please help, hopefully I can get this sorted out.

Thanks guys.

Disable C1E and both C3 and C6 Reports. With 8GB of RAM installed, you'll need to bump VCCIO voltage above the default of 1.05V. What are the exact spec's of your RAM, including timings and voltage at DDR3-1600, and what's the current DRAM frequency? And you can probably get by w/ 1.30V or less Vcore at a 4.5GHz clock... 1.35V is too high.

I ran prime95 blend for 9 hours. After about an hour, one of the workers (#4) failed due to rounding error. The other three workers were fine.

Click to expand...

Once a worker fails there's no need to continue running Prime, since your overclock is unstable at that point anyway.

Suggestion - as mentioned above, tweak VCCIO up to 1.1 or 1.15... see if that helps. I'd also raise CPU current capability a bit.

You're already doing better than my extra spicy chip does at those settings. Mine will not do 4.5 at that vcore... it likes 1.40V to be stable. I have the pro version of that board, and same memory only double the amount. I'm still trying to stabilize 4.6GHz. Down to a BSOD at random once every 12 hours on average, so I'm getting there.

For mine, I have to have LLC set to ultra high (due to high vcore), PLL overvolt disabled, memory settings using the XMP profile with 1.55V, VRM frequency manual at 350, duty control extreme, phase control extreme, and set CPU current to the highest setting. From there, 1.4V core, 1.7V PLL, and 1.1V VCCIO is what it takes to stabilize this chip at 4.5GHz. All CPU power management options enabled.

Remember, though, mine's an ES that seems to be a little power hungry as an overclocker. Yours should manage 4.5 a fair bit easier. Try the little things first, like tweaking VCCIO up a bit, and manually setting PLL voltage to 1.75 or so. Don't trust the auto setting to do the best job

Hope this helps... just got back into overclocking after years of being lazy.

redduc900 said:

Disable C1E and both C3 and C6 Reports. With 8GB of RAM installed, you'll need to bump VCCIO voltage above the default of 1.05V. What are the exact spec's of your RAM, including timings and voltage at DDR3-1600, and what's the current DRAM frequency? And you can probably get by w/ 1.30V or less Vcore at a 4.5GHz clock... 1.35V is too high.

Once a worker fails there's no need to continue running Prime, since your overclock is unstable at that point anyway.

Click to expand...

9-9-9-24-2N 1.5v is my RAM timing. I'll go back in and disable C1E and C3/C6 reports. I'll raise VCCIO as well. Anything else I need to do? There has been some debate if I should run Small FFT or Blend mode on prime95. What do you guys suggest? Thanks so much.

EDIT: Raised VCCIO to 1.1v, brought CPU Current to 140% (max setting) and dropped Vcore back to 1.35v. Disabled C3/C6 and C1E. Running Small FFT now.

Worker #4 FATAL ERROR rounding after one hour. What now? :/

You have two threads going essentially (in the 4.5ghz oc thread)...

Now I see why they said to raise VCCio for the ram, not the CPU.

The 2500k is a voltage gobbler compared to the 2600k. Like OW knows, 1.4v around 4.5Ghz is common(ish - a bit on the high side). Like I mentioned in the other thread, run small fft to make sure we know its the CPU, and raise the voltage a bit.

When you raised the voltage does the time before failure increase?

If you are looking for CPU stability, run SMALL FFT. Like I mentioned in the other thread, BLEND tests a lot of memory.. I just dont think we are there yet... But theocnoob likes to do blend first.. there isnt a wrong answer, just a more efficient method. If you are stable, the running blend first makes sense. If you dont know if your CPU is stable, SMALL FFT makes more sense.

EarthDog said:

You have two threads going essentially (in the 4.5ghz oc thread)...

Now I see why they said to raise VCCio for the ram, not the CPU.

The 2500k is a voltage gobbler compared to the 2600k. Like OW knows, 1.4v around 4.5Ghz is common(ish - a bit on the high side). Like I mentioned in the other thread, run small fft to make sure we know its the CPU, and raise the voltage a bit.

When you raised the voltage does the time before failure increase?

If you are looking for CPU stability, run SMALL FFT. Like I mentioned in the other thread, BLEND tests a lot of memory.. I just dont think we are there yet... But theocnoob likes to do blend first.. there isnt a wrong answer, just a more efficient method. If you are stable, the running blend first makes sense. If you dont know if your CPU is stable, SMALL FFT makes more sense.

Click to expand...

Yeah, I'm failing Small FFT after one hour or so. No idea what to do now :/

zackj3 said:

Yeah, I'm failing Small FFT after one hour or so. No idea what to do now :/

Click to expand...

You've ruled out VCCIO, so give vcore a bump to maybe 1.375V next. See if that changes anything. Temps look ok, so no worries there yet. Mine wouldn't stabilize at 4.5 with 1.375V (almost, but not quite), but that's not to say yours won't.

I think my highest core load temp at my stable 4.5 overclock was something like 71 degrees. Just try and keep it below 80.

EarthDog said:

Like OW knows, 1.4v around 4.5Ghz is common(ish - a bit on the high side).

Click to expand...

Well, to be fair I was going in 0.025V steps... didn't really try to find the optimal voltage for 4.5 due to being obsessed with getting 4.6 stable

Oklahoma Wolf said:

You've ruled out VCCIO, so give vcore a bump to maybe 1.375V next. See if that changes anything. Temps look ok, so no worries there yet. Mine wouldn't stabilize at 4.5 with 1.375V (almost, but not quite), but that's not to say yours won't.

I think my highest core load temp at my stable 4.5 overclock was something like 71 degrees. Just try and keep it below 80.

Click to expand...

Do I still keep VCCIO at 1.1v? What about CPU Current? I currently have it at 140% (Max Setting). Should duty control and phase control both be set to extreme? PLL? VRM Frequency? Not sure what to have all of these at.

I'd leave them where they are for now. Won't hurt anything. You can always play with those later when you've gotten things more stable. I didn't really see much difference moving that CPU current option around, if I'm being honest. Maybe because I'm not pushing my chip to 5GHz... yet. It's there to limit the CPU's current draw.

I get less stability if I try to drop my VCCIO any lower than 1.1, but then again I'm running two sets of Ripjaws, and they're not totally the same parts. One set's XMP certified, the other's merely "compatible." They both have the same timing specs. Anyway, YMMV.

Just to give a second reference to what Earthdog said regarding 2500k and voltage, the 2500k in sig rig #1 absolutely will not mantain stability at 4.6 without 1.38 vcore no matter what I do, and I have twiddled for hours. If it drops below 1.38v at load I will bsod with a 101 error as surely as death and taxes.

That's not saying not to try, rather that if you find yourself bumping volts you're in a lot of good company. And I haven't seen anyone here thats a MOARVCORE guy

Ran prime95 Small FFT for close to 15 hours. No failed tests. Here are screen shots. What is next? Blend?







As you can see, I have VCCIO at 1.1v, and CPU voltage at 1.365v. That seems so high compared to what everyone else gets.

Are my temperatures okay? What is next, run prime95 blend?

Temps are ok. Glad to see an improvement. Unfortunately, there are no guarantees with overclocking... it takes what it takes.

Finally got mine stable at 4.6, it seems, without going past 1.425V. PLL overvolt enabled is what did it. Letting it go a bit longer before I try for 4.7 though... need to be sure my random idle BSODs are gone. I'm going as high as I can with this chip without passing 80 degrees.

Next? Looks like you're totally stable, to me. Will let others chime in there. I'm rusty at this, remember.

Oklahoma Wolf said:

Temps are ok. Glad to see an improvement. Unfortunately, there are no guarantees with overclocking... it takes what it takes.

Finally got mine stable at 4.6, it seems, without going past 1.425V. PLL overvolt enabled is what did it. Letting it go a bit longer before I try for 4.7 though... need to be sure my random idle BSODs are gone. I'm going as high as I can with this chip without passing 80 degrees.

Click to expand...

Thanks for the quick response, but I'm not sure what to do next. Should I change any of the BIOS settings? Run Blend for a while? This is my first go at overclocking and I'm pretty ignorant when it comes to all of this.

Hope yours works for you.

You posted while I was ninja editing

As I said in the edit, you look stable from here. I'm rusty at the overclocking, so I'll let others comment. I'd be saving that to a BIOS profile and trying for 4.6. Onwards and upwards.

Ran prime95 blend for 6 hours with no failed tests after running prime95 Self FFT for 15 hours with no failed tests. Here is a screen shot.



Curious why VCORE reads a max of 1.33 when I have it set to 1.365 in BIOS. What can I do now? Any settings to adjust in BIOS? Go for higher? Try to get it stable at <1.35v?

Higher? Thats not for us to decide?!!!

That voltage difference is called Vdroop. This is prevent by using LLC. However, you dont need it. Seems you are stable at 1.33v... NICE!

EarthDog said:

Higher? Thats not for us to decide?!!!

That voltage difference is called Vdroop. This is prevent by using LLC. However, you dont need it. Seems you are stable at 1.33v... NICE!

Click to expand...

If I'm stable at 1.33, why can I not go in and set VCORE to 1.33 instead of 1.365?

I just gamed for a little bit and my max VCORE according to CPUID HM was 1.35.

Because when you lower it, it will then lower again in windows. 1.365 bios = 1.33 load. 1.33 bios = 1.3 load.

So since I was failing tests at 1.35v earlier, it is safe to say that I will not be able to ever obtain a stable 4.5ghz overclock @ 1.35v?