Overclocking weirdness, is "burn-in" real?

[MagickMan]

I have a 5930k and it's a fair overclocker, 4.5GHz @1.3v, but today I thought I'd give 4.6+ another shot. Well, after just changing the multiplier to 46, not adjusting anything else, the system is running fine.

Boots up properly.
Run Realbench benchmark, pass.
Run Realbench 15 minutes, pass.
Realbench 1 hour, pass.
2 hours, pass. No hiccups, no errors. Temps stay in low 70s under full load.

WTF? This machine, in particular this CPU, would in no way pass even the Realbench benchmark at 4.6GHz before (and sometimes even apps would crash), so I settled on 4.5 for the time being, until I had more time to tinker. But now it's running like a champ (still won't boot at 4.7, not even @1.35v). Could it be some type of burn-in effect, either with the chip or the motherboard?

 

[i7Baby]

Maybe last time you increased voltage.

Maybe this time the CPU cooler thermal paste has had time to spread better and is cooling the CPU a touch better.

 

[jji7skyline]

Arctic Silver 5 claims to have a 200 hours burn in time or something.

More likely it's because of lower ambient temperatures, case temperatures, and/or voltage settings.

 

[YBS1]

Updated the BIOS by chance? Sometimes certain BIOS versions are notably better clockers.

 

[MagickMan]

Nope, no voltage changes, and temps have never been a problem (I use IC Diamond compound). It's literally set at the same voltage settings it was when it previously failed.

Now it's gone through another 4 more hours of Realbench. Huh... oh well.

 

[Idontcare]

Its a real-life user machine, right? Meaning you have countless apps, including your OS, updating in the background over the passing months and weeks...any of which could have been coincidentally the rate limiting stability issue in the past.

When it comes to real-world end-user systems,software stability over time is THE biggest unknown for the hardware guys which is why they must over-engineer and guardband the heck out of any degree of drift that will transpire in the hardware itself. And from that over-engineering you (we consumers) get the opportunity to OC.

So your chip is a corner condition for which your specific software suite just so happens to, now that it has been patched X number of ways, be amenable to an extra 100MHz of OC'ing. Don't over-analyze it, just go with it.

 

[JimmiG]

Its a real-life user machine, right? Meaning you have countless apps, including your OS, updating in the background over the passing months and weeks...any of which could have been coincidentally the rate limiting stability issue in the past.

When it comes to real-world end-user systems,software stability over time is THE biggest unknown for the hardware guys which is why they must over-engineer and guardband the heck out of any degree of drift that will transpire in the hardware itself. And from that over-engineering you (we consumers) get the opportunity to OC.

So your chip is a corner condition for which your specific software suite just so happens to, now that it has been patched X number of ways, be amenable to an extra 100MHz of OC'ing. Don't over-analyze it, just go with it.

Yeah, weird stuff can happen at the edge of how far your chip will OC. I've had chips pass all sorts of stress tests, only to BSOD when playing a Youtube clip or running Minecraft or something.

 

[MagickMan]

Its a real-life user machine, right? Meaning you have countless apps, including your OS, updating in the background over the passing months and weeks...any of which could have been coincidentally the rate limiting stability issue in the past.

When it comes to real-world end-user systems,software stability over time is THE biggest unknown for the hardware guys which is why they must over-engineer and guardband the heck out of any degree of drift that will transpire in the hardware itself. And from that over-engineering you (we consumers) get the opportunity to OC.

So your chip is a corner condition for which your specific software suite just so happens to, now that it has been patched X number of ways, be amenable to an extra 100MHz of OC'ing. Don't over-analyze it, just go with it.

I keep a fairly clean machine, a minimum of superfluous apps, but it is my regular gaming machine. The shift from complete instability, with anything, to complete stability over a span of 3 months is just... strange, and doesn't appear to me to be software related. If anything, software/OS degradation should make it less stable. I did read a comment from an Asus engineer that motherboard voltage regulators can often become a more efficient and accurate over time, so I suppose that could be a possibility.

 

[MagickMan]

3 hours of IBT followed by 6 hours of Witcher 3, I'm going to go ahead and call it stable. Strange.