Overclock stable in Prime95 but fails IBT

[ArjenVdv 190]

Hi,

 

I run 4.7 GHz at 1.39V for nearly a year now but just decided to do another stress test today because stability isn't forever. Just like last year I did a 12 hour small FFT Prime95 torture test and it passed without any errors. Max core temps was 71C. For a change, I decided to run Intel Burn Test for maximum stress testing. I opened task manager, CPU-Z and HardwareMonitor, and used the default settings for IBT (10 runs + standard stress level). I clicked Start and surprisingly, after only 5 runs my PC suddenly froze up. No BSOD, just a freeze, and I had to shut down holding the power button of my PC. Tried again raising the VCore, until ending up with 1.47V, but everytime after 5 runs or so my PC froze up again. So my VCore must be fine.

 

I'm a fairly experienced overclocker and the settings I'm using seem fine to me. The fact my PC has been running without a single problem for nearly a year now and passing Prime95 for 12 hours, but failing IBT after half a minute, leaves me completely puzzled.

 

Here are pictures of my BIOS settings, I experimented with all of them and I believe these are the optimal settings. Offset voltage of +0.040 gives me on 1.39V on load by the way.

 

Any ideas on why my PC freezes on IBT so quickly?

 

Thanks...

 

 

 

 

[Ben Cap 89]

I've never personally used IBT for stress testing. - It loads the CPU with an unrealistic amount of load.... There is no way that you would *ever* come close to that kind of load in *any* kind of scenario in FSX.

 

P95 is the way to go. B)

[PieEater 10]

Both Prime and IBT stress the CPU in different ways, I have had my PC pass IBT but fail Prime after 2 hours. Neither method is a guarentee of system stability for anything other than running Prime or IBT, i.e. FSX will never stress your CPU like IBT so although your PC may not be IBT stable it may well be stable for FSX and most other applications.

 

IBT is a bit more agressive than prime in that it does generate a lot of heat and very quickly, if your system passed IBT a year ago but doesn't now it could be heat related, have you cleaned out your PC recently? Any dust buildup can adversely effect cooling so it's worthwhile giving your case and componants a good spring clean and trying again, even with dust filters my CPU and GPU coolers still suffer with trapped dust after a few months. I generally use an artists paint brush to clean individual componants, make sure the motherboard is clean and remove dust from fan blades and inbetween heatsink fins, I remove my graphics card at the same time to give this a good clean.

[martin-w 3,234]

If it's stable in everyday use, and Prime, then I say don't worry.

 

As said by others above, IBT is vicious and not comparable to anything your CPU would encounter, day to day.

 

It doesn't matter how stable your system is, it may be totally stable in IBT, but if I were a skilled programmer I could code a stress test that would still bring your system to it's knees... so it makes no sense to keep increasing voltage to accommodate unrealistically demanding stress tests.

 

If it's stable day to day, don't worry.

 

CPU's can become unstable and require more voltage due to degradation, but I don't believe that's the issue for you. CPU lifespan is measured in thousands of hours, don't think degradation is something you should entertain as a cause. Especially as you said you ran IBT "for a change" implying you didn't originally. It may not have passed IBT if you had ran IBT a year ago.

[ArjenVdv 190]

Ok, but the thing is, all the runs it does it ends up with the same code on the right hand pane which is an indicator of stability as far as I've heard. And my PC doesn't BSOD, it just locks up, freezes after 30 seconds of IBT. Most people can easily run IBT, and even though it's rock stable in Prime95, the fact that it locks up in IBT is indicating something is not entirely right.

 

if your system passed IBT a year ago but doesn't now it could be heat related

 

I have never run IBT before, I always used Prime95 only. Yesterday was the first time ever I ran IBT.

[martin-w 3,234]

Temps in IBT?

 

I think it unlikely that 4.7 GHz at 1.39V for a year would degrade the CPU enough to require substantially higher voltage.

 

Am I right in saying you "didn't" run IBT a year ago, when you first stress tested?

 

So it may have failed in IBT, one year ago, if you had stressed with IBT back then.

[ArjenVdv 190]

Temps in IBT?

 

Max core temp in the high 70's, around 77-78C. Max Package temp in the low 80's. Max temps in Prime95 are in the high 60's. And this is on a $30 air cooler.

 

Temps in IBT?

 

Max core temp in the high 70's, around 77-78C. Max Package temp in the low 80's. Max temps in Prime95 are in the high 60's. And this is on a $30 air cooler.

 

Temps in IBT?

 

Max core temp in the high 70's, around 77-78C. Max Package temp in the low 80's. Max temps in Prime95 are in the high 60's. And this is on a $30 air cooler.

 

Temps in IBT?

 

Max core temp in the high 70's, around 77-78C. Max Package temp in the low 80's. Max temps in Prime95 are in the high 60's. And this is on a $30 air cooler.

[martin-w 3,234]

Is this temp measured in Real Temp? With the TJMax set correctly in Real Temp? And temp verified as correct by something else?

 

Thats very low temp for a $30 air cooler. I wonder if your temps are reading accurately.

[ArjenVdv 190]

Is this temp measured in Real Temp? With the TJMax set correctly in Real Temp? And temp verified as correct by something else?

 

Thats very low temp for a $30 air cooler. I wonder if your temps are reading accurately.

 

I measure it in HardwareMonitor and CoreTemp 1.0. CoreTemp shows temps that are around 3C lower. According to CoreTemp, TJ max was set to 98C.

[ArjenVdv 190]

Ok, this is a bit odd... I just fired up IBT again but this time changed stress level from Standard to Maximum and now my PC has been running rock solid for 10 minutes? *Knock on wood*. Max core temp in HWM is 86C and 90C in CoreTemp 1.0. For short testing I think this wouldn't do much harm?

[PieEater 10]

When IBT fails the system freezes and you have to hard reset as you describe, this is normal.

[martin-w 3,234]

Ok, this is a bit odd... I just fired up IBT again but this time changed stress level from Standard to Maximum and now my PC has been running rock solid for 10 minutes? *Knock on wood*. Max core temp in HWM is 86C and 90C in CoreTemp 1.0. For short testing I think this wouldn't do much harm?

 

Strange.

 

May be just an IBT glitch.

 

I know for my Ivy Bridge, both Intel and Asus advised against it's use. It's not validated for Ivy.

 

Those temps look high. I wouldn't sustain that too long, it's not far off TJMax and throttling. I mean it will shut down if it gets any higher.

[ArjenVdv 190]

Awww, after running it for 20 minutes it froze after the 9th (out of 10) run. Still, I think my PC did good. Maximum stress level, temps were high, and it managed to keep stable for a fairly long period. I think my PC must be pretty solid

 

@Martin. HWM does measure core temps yes, for each core seperately. The tempatures are a bit different from CoreTemp though. Max core temp in HWM was 86C, and in CoreTemp it was 91C. In Prime95 though, temps are between 15-20C lower so I will be fine in general use. It idles at only 25C.

And to answer your question, HWM also shows the Package temp right under the core temps, it was around 5C higher than the max core temp (91C). The only other CPU related tempature it showed was CPUTIN, which is always at 60C for some reason.

[martin-w 3,234]

Did you test with IBT a year ago, when you originally stress tested? If not, I wouldn't worry about it if I were you. If you have been stable in Prime95, and for everyday use, for over a year, then you don't have a problem. The true test of an overclock, is long term everyday use. Because that's what we require our overclocks for.

 

If on the other hand, you did test a year ago in IBT and it passed, and now suddenly it doesn't, that would be more of a concern.

 

You can expect to require higher voltage to be stable in IBT.

 

IBT generates more heat, and when you generate more heat, the CPU requires more voltage.

 

The voltage you are useing equates to stability in Prime and everyday use. And you say that in that regard, you are stable.

 

Some may disagree with me, but I would say stress tests are both a good thing and a bad thing. They are a good thing, in that they enable us to quickly test our systems for stability. But then "extreme" stress tests, like IBT, that generate a lot of heat can be a bad thing, simply because they encourage us to use a higher voltage than we actually require for long term everyday use.

[ArjenVdv 190]

Ok great... after having been stable for a year I just encountered a BSOD in mid flight with the 124 error, not enough voltage I guess. Could the 15 minutes of IBT of this afternoon already have degraded my CPU while it ran at nearly 90C? I hope not...