[Kanashimu]

I've recently been concerned by the potential of damage and degradation from heavily overclocked and overvolted CPUs being subjected to multi hour stress tests. My 2500k has seen serious degradation over time, and requires a lot more vCore to run at a far lower clock. I personally try to go as high as my CPU possibly can, but the combination of high voltages and high temps are deadly.

Is there are more ideal way to stress test a CPU? Is there such thing as an effective stress test without pushing temps very high? For example, if I use prime95 and I set FFT size bounds to not have any of the small FFTs, I'll generate far less heat, but will I still be running a useful stress test?

 

[billbartuska]

Quote:

Originally Posted by Kanashimu

For example, if I use prime95 and I set FFT size bounds to not have any of the small FFTs, I'll generate far less heat, but will I still be running a useful stress test?

No you wouldn't. The whole purpose of stress tenting is to be sure your CPU/GPU will be stable when performing at it's maximum.

If you test at a reduced stress then you have no idea whether you CPU or GPU will be stable at any higher stresses than that, so why bother stress testing at all?

Then there's the school of thought that says forget stress testing altogether to just use the computer as you normally would and see if it performs satisfactorily. If it does then you're good.