Optimum slack time

Some quality chaps at the University of Arizona put together a report calculating how long you should slack for. The Effects of Moore’s Law and Slacking on Large Computations (beware PDF) Essentially, the idea here is that if you’re running a long computation, you might be better to sit on your hands for a bit and wait for computing speed to get faster before starting it; if computers get faster enough quickly enough then you might end up finishing before you would have if you’d started straight away. Or, in one of the more memorable phrases from the report

[Y]ou could start a computation now, calculate for 40 months, and get a certain amount of work done. Alternately, you could go to the beach for 2 years, then come back and buy a new computer and compute for a year, and get the same amount of work done.

I wasn’t kidding about it being for long computations, though. t~0~(s), the time for which it’s always worth starting now rather than waiting, is 26 months. If your calculation is going to take less than two years then you might as well start now, and (and I don’t think that this will come as a surprise to anyone) I personally don’t have any computations that will take that long currently outstanding on my to-do list. As the report has it

Note that the size of the calculation does not vanish as s → 0, ie. there is a minimum calculation for which it is ever worth it to slack. This is reassuring, since otherwise it would always be worth it to wait and we would never get anything done.

Genius. Unutterable genius. Well done Chris Gottbrath, Jeremy Bailin, Casey Meakin, Todd Thompson, and J.J. Charfman.

More in the discussion (powered by webmentions)

  • (no mentions, yet.)