On Sat, 06 Jul 2013 03:05:30 +0000, Steven D'Aprano wrote: > That doesn't explain how you time it, only that you have a loop executing > 100 times. Are you using time.time, or time.clock? (I trust you're not > measuring times by hand with a stop watch.) > > I expect you're probably doing something like this: > > start = time.time()
I have been using time.process_time >> If the timing version, which executes function "Solve" one hundred >> times, runs about 80-100 seconds without a significant variation, then >> taking the mean is mathematically correct. > For longer running code, like this, you might also like this: > http://code.activestate.com/recipes/577896/ Thanks for the pointer. > If the best you can say is it takes "80-100 seconds", that's pretty > significant variation, of the order of 20%. That's not a variation of a SINGLE variant. One variant takes 80 seconds and the other variant to be compared with takes 100 seconds. > > In this case, with times of the order of a second per loop, it may be > reasonable to say "in this specific case, the error is too small to care > about", or "I just don't care about the error, since it will be about the > same for different variations of my solve function". But in that case, > why bother looping 100 times? > > >> I can't take the minimum >> since I don't measure the time a single call takes. > > Then perhaps you should. Many thanks, Helmut -- http://mail.python.org/mailman/listinfo/python-list
