CPU Stability: different results from different programs

[Nemesis1598]

So I've got an FX-6300 and today I decided to try and push the overclock a bit further. I bumped it up to 4.5Ghz and started up the stress tests.

I ran Aida64 for 2 hours with no issues, however when I did the small FTT test in Prime95 I encountered an error.

 

Would this be considered unstable? I know Linus uses Aida64 for all his testing, and if I recall, he mentioned in his Skylake OC guide that newer versions of Prime95 can be unreliable?

 

(And don't worry, I usually run the test at least overnight, not for just 2 hours)

[Altecice]

Yup that's unstable, I have had OC's sit in Aida64 for 24h and then start a game and instantly crash.

[Imakuni]

So I've got an FX-6300 and today I decided to try and push the overclock a bit further. I bumped it up to 4.5Ghz and started up the stress tests.

I ran Aida64 for 2 hours with no issues, however when I did the small FTT test in Prime95 I encountered an error.

 

Would this be considered unstable? I know Linus uses Aida64 for all his testing, and if I recall, he mentioned in his Skylake OC guide that newer versions of Prime95 can be unreliable?

 

(And don't worry, I usually run the test at least overnight, not for just 2 hours)

If it crashes once = unstable.

 

You can decide "okay, I'm never going to stress my CPU that much anyway" and stick with the OC regardless of it crashing. But by definition, if it crashes = unstable. Boost voltages or reduce clocks to try and fix it.

[BrightCandle]

CPU's and the software that run on them are not supposed to crash or produce the wrong results. The expectation of all software is that the hardware is reliable. If it ever crashes or miscalculates then its not functioning as designed.