CPU stable, then unstable?

[Stiffmeistro]

Hi nerds!

 

I recently overclocked my FX-4100 to 4,6 GHz and used Prime95 for 5,5 hours to stability test it, and it was ok. (No warnings or errors or anything.)

I then tried to Prime95 again, this night, and it only did it for 35 minutes before prime95 said there was a error of some kind. (0 errors 100 warnings, worker stopped)

 

How come? I have not changed any bios settings.

 

Asus Sabertooth 900fx R2.0

FX-4100

Corsair CX750M

[Levent]

did you set your voltages manually? if its on Auto your motherboard might be overvolting it

[Stav Kats]

5.5 hours is not much to make sure that your system is stable. You should AT LEAST run Prime95 for 8 hours. Although, I would recommend running it for 24 hours if possible.

[TheCarioca]

Well, for me, 5.5 hours in prime95 doesn't mean it is stable. I run it for at least 24 hours. Try bumping up the voltage a bit and consider it stable only if it passes 24 hours. If you're lazy and still want to overclock it more, try 12 hours tests until it gets to the point you want and then tweak it until it passes the 24 hours in prime95.

 

These are my two cents about stability. About the tweaking per se, I'm shure someone else can help you more with that.

[Stiffmeistro]

I manually set the voltages, yes.

I'll try to up the voltage abit more then.

 

Thanks

[Opsilon]

Dude Prime95 is only good if you do it for 8 h, but Linus even said 24-48 hours of testing, depending on the chip, i would not go overkill with that chip, sure 4.6 is obtainable, but that does not mean every chip can do that, maybe keep it at 4.4, would be enough, you will have a small decrease in performance vs 4.6, but you will be able to keep it sable, and i only benchmarks will see the 200 mhz increase as a "significant" boost

[Stiffmeistro]

I tried doing 20 runs of Intel Burn Test maximum, and it seems stable.

 

the results were 3.292699e-002 in all 20 test's.. Isent that supposed to be stable, and if its unstable, there will be another number?

[RDilux]

I tried doing 20 runs of Intel Burn Test maximum, and it seems stable.

 

the results were 3.292699e-002 in all 20 test's.. Isent that supposed to be stable, and if its unstable, there will be another number?

doesnt matter 8-25 hour prime95 on Small FFT or Big FFT bump up the vcore if not stable