Hi, I'm wondering if anyone could make a tutorial about benchmarking Sage methods. I am personally not much of a speed fanatic, but when I am refactoring other people's code I'm trying to make sure it doesn't slow down as the result and this does require me to run timings. I'm aware of the basics of %time and %timeit, but here are some questions I haven't directly seen answered:
- How to usefully measure the runtime of a method that involves subprocesses (GAP, symmetrica, whatever)? I don't think it appears in the output of %timeit. - Why does %time often gives a longer time (both CPU and wall) than %timeit, even if there is nothing to output to IO? - How do %time and %timeit deal with the cache? Am I seeing it right that any instructions following a %timeit have no effect on the state (they get executed in a kind of sandbox?) while those following a %time are processed like any other instructions? - What is the most accurate way to measure changes to startup time, other than starting Sage many times and measuring wall times? - How can I find out how much a change in some class has slowed down / sped up methods in other classes? I could probably run doctests, but I'm not sure how realistic they are (some of them might even be nondeterministic, causing further noise), and I would probably have to write some scripts to process the output. Are there good tools that predict the most likely places where a code change can cause consequences? - Memory is another issue: how can I see if my cache runs the danger of blowing up? Thanks a lot, Darij -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/groups/opt_out.
