Trying to get my 2500k to a stable overclock

[apav]

----------------------------
IntelBurnTest v2.53
Created by AgentGOD
----------------------------

Processor: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Clock Speed: 3.30 GHz
Active Physical Cores: 4
Total System Memory: 8168 MB

Stress Level: Maximum (7144 MB)
Testing started on 5/28/2012 4:56:04 PM
Time (s) Speed (GFlops) Result
[16:59:08] 151.589 119.2417 2.949493e-002
[17:02:12] 151.107 119.6220 2.949493e-002
[17:05:15] 151.762 119.1053 2.949493e-002
[17:08:16] 150.272 120.2869 2.949493e-002
[17:11:17] 149.812 120.6561 2.949493e-002
Testing ended on 5/28/2012 5:11:17 PM
Test Result: Success.
----------------------------
I don't know why it says it's at 3.3Ghz, because CPUID reads 4.5, maybe it's talking about stock, but I know it's overclocked.

Anyway, it survived the burn test, would you say it's stable then? Temps hit on average mid-high 80's, 89C max, but I know the point of this test is to really crank up the heat. And I don't nearly get as much doing other things. I idle in the 40's.

 

[apav]

I changed back VCCIO and CPU PLL back to auto. I kept LLC, Phase Control, and Duty Control to extreme though. I'm trying to figure out if I should revert the last to to auto or LLC to a lesser setting.

Either way, it survived Intel Burn Test without crashing (at 1.35V), so I know something is right.

 

[Automata]

You are going to want to run more than 5 iterations of that test.

 

[apav]

You are going to want to run more than 5 iterations of that test.

That's how many Theocnoob said to run in his guide. How many would you recommend? I just don't want it running in the high 80C's for a long time. 15 minutes was already driving me crazy (I thought it was gonna overheat!).

 

[Automata]

80c is not going to hurt that chip, even if you ran it all the time. These are good past 90c.

5 passes is a good quick test of stability where you just want a quick test, such as when you are finding your max overclock speed. I actually max out the iterations and just let it run 24 hours. If you were to let it go for two hours, I think that is a pretty good indication that it will probably work in whatever programs you will be running.

Verbose version:
Since there is no way to know whether a chip is fully stable (short of running stability tests until it dies, which isn't very useful), you can only prove it is a mathematically insignificant chance. The longer you can run the test and the more stressful the test is (including higher temperatures), the less chance you will have an issue later on. When I was doing my 24 torture test with my current settings, the ambient temperature of the room was much higher than I would see during normal usage. Since I will also not be stressing the processor anywhere near what LinX would do, there is a very small chance that I will run into issues with day-to-day programs.

Really, it comes down to how stable you want it. Benchmarkers only need the system to be stable for a test or two. Gamers need it to run their games. Coders need it to compile the code perfectly every time. You just need to find where on this spectrum you are.

 

[apav]

55 minutes of Linx so far with these settings and it hasn't crashed so far... how long do I need to stress test it for?

It isn't crashing because I have LLC on extreme. My voltage is mostly 1.36 ( but does hit 1.344 occasionally). I must have one of those unlucky chips that will crash at 4.5ghz if I go below 1.35V, or even AT 1.35V.

 

[Automata]

48 minutes of Linx so far with these settings and it hasn't crashed so far... how long do I need to stress test it for?

As I mentioned in my post just before yours, how long you let it run is a function of how stable you need the system to be. If you are gaming, 2 hours should be fine.

 

[apav]

As I mentioned in my post just before yours, how long you let it run is a function of how stable you need the system to be. If you are gaming, 2 hours should be fine.

Great! I'll let it run for another hour, then I have to decide if I want to change the LLC back to high and change the Phase Control and Duty control back to auto.

I guess I'll also try lowering the voltage by a little. But I really think setting LLC to extreme is what's making me stable. Because I'm getting more than 1.35, which is what I set the manual voltage to.