Super PI

[BTRY B 529th FA BN]

Any reason why SuperPi would work then the next time fail in the same boot?

 

[Idontcare]

SuperPi is supposed to be sensitive to memory and NB/IMCFSB/HT speeds and stability.

If it works intermittently than it suggests that you have a component in that loop which is right on the edge of stability in terms of heat and voltage, the usage pattern of SuperPi itself might be just enough to heat up the components but not enough to cause the program to error.

Then if you relaunch another superpi bench before the now (pre)heated components had a chance to cool back down to their typical near-idle temperatures then the newly relaunched SuperPi bench will heat up the components all the more and the weakest component will finally be placed into a thermal/voltage environment for which it is unstable and the bench crashes.

(edit: the above sentence was nearly unintelligible in its original form, I edited it for clarity and smacked myself with a wet-fish for good measure)

Same reason people like to loop linpack, prime95, memtest, etc. Unfortunately Superpi doesn't really have a loop feature built-in, and because of the nature of the iterative equations employed to compute digits of Pi a 32M superpi run is not the same as looping a 1M superpi run 32 times (because its not the same data/values being used in the iterations, so workload and heat output changes as well as memory footprint).

If your rig fails to run superpi consistently then I'd recommend checking prime95 large FFT stability for a good 12hr run or so.

 

[BTRY B 529th FA BN]

Awesome! Thanks IDC - :beer: