Sign in with
Sign up | Sign in
Your question
Solved

Unstable overclock, not sure why. Any advice is appreciated.

Tags:
  • Overclocking
  • Cooling
  • Intel i5
Last response: in Overclocking
Share
June 9, 2014 5:42:19 PM

So I got a GIGANTIC Noctua air cooler for my aging i5 750 CPU and two GIGANTIC Arctic Accelero air coolers for my aging crossfired HD 6770 GPUs thinking that if I OC'd the hell out of them I could get another year before I have to upgrade my hardware.

This screenshot should tell you pretty much everything you need to know;

http://i.imgur.com/EdqVqe0.jpg

Using 2011's Tomb Raider and Bioshock Infinite as benchmarks (Highest settings) I got a 30% FPS boost with these settings and I'm running 30 degrees cooler to boot!

Trouble is, my machine keeps randomly locking up and/or shutting down. It'll run Prime95 for over an hour without so much as a peep and the Thermal Margin never dipping below 70F, but if I just let it sit, it will fail. . . Eventually. Sometimes it takes half an hour, sometimes it takes 2 days, but after X amount of time I'll find the computer either on but completely locked up, or off completely. If it's off then I get a warning beep when I start it back up and a screen that tells me the computer was automatically shut down in response to a "thermal event."

Thermal Event!! My GPU's are idling at barely 100F, and just barely kiss 130F running Furmark over and over and over again. The CPU idles with a Thermal margin in the mid 150'sF and an hour of Prime95 brings that down to 70F. (LOVE that Noctua air block, btw! :) )

My point, I suppose, is that this is very difficult for me to troubleshoot. I zeroed in on these settings very carefully and conservatively, and between Prime95 and Furmark I thought I had it nailed down. But these random sometimes-lock-ups and sometimes-shut-downs that occur anywhere from 30 to several thousand minutes apart are making it tough for me to tell what exactly I've got wrong and need to change. If the damn thing would just refuse to post or fail to boot, then fine, at least I'd know to undo what I just tried, but when it runs fine for 36 hours. . .

I suspect the culprit is my vtt, (this Bios calls it "uncore") because Intel specifically says not to exceed 1.21v with this chip, but with this 2-ton chunk of copper bolted to the CPU I figured I had a little wiggle room with that number. I also bumped my core voltage up a few times after the first few freezes hoping that would stop these incidents from happening. It was originally at 1.475v when I found my "stable" OC. Has this 0.0375v increase helped? Not helped?

Anyway, looking at that screen, does anyone more knowledgeable than I see something obviously out of whack that I should focus on? Any tips for nailing this down?

Thanks for any help or advice!

Specs:
Mobo: Intel DP55WG
CPU: Intel i5 750
GPU: 2x Radeon HD 6770
RAM: 8Gb (4x2Gb) 1600MHz DDR3 9-9-9-24
Windows 7 Ultimate 64bit




More about : unstable overclock advice appreciated

June 9, 2014 5:53:01 PM

Overclocking is a gamble in it's own right. Really what you're doing is forcing the processor to run at a speed that it was never designed for, and therefore under a stress that it was never tested to handle. I don't believe heat is your problem as your temps are normal. Really, it may simply be that your processor is unstable running at those speeds and will lock up randomly. Try dialing down the overclock a little, about half of what you upped it and see if the problem persists. Your processor might like that speed better and that might be all you can push out of it.
m
0
l

Best solution

a b K Overclocking
June 9, 2014 6:08:20 PM

My money would be on the mobo. Intel doesn't build a super overclocker board like Asus or Gigabyte, and as such I think, especially with a 1.21v cap, that you are probably overheating the VRMs. When using prime95, your cooler is running faster and harder, creating lots of 'bleed' airflow, which helps cool the VRMs which sit right above the CPU socket. At idle, or low CPU usage, the fans are spinning super slow, so very little 'bleed' air is cooling the VRMs, yet you are still pushing raised voltage through them, so they overheat, and the CPU goes bunk.

3 fixes for the above.
1. Install a dedicated fan to blow constant air over the VRMs area.
2. Turn your voltage down considerably, and the OC with it.
3. Buy a new cpu/mobo since its only a matter of time before either the VRMs burn up, or bad voltage fries the CPU.
Share
June 9, 2014 6:17:09 PM

Well Poo. . .

That's 2 votes for dialing down my vtt, and by association, the bclck.

I really want to keep that 200MHz bclck because then with a 4X multiplier on the RAM it's running right where it should be. It makes the math so much easier.

I guess I'll start over, running Prime95 for a few hours instead of a few minutes between attempts and see where that lands me.

Thanks for the tips, guys.

C
m
0
l
!