Closed nishihatapalmer closed 5 months ago
Good to measure variability with a different number of runs, and quantify it.
On the other hand, if we start measuring different processes and use a larger text size by default, 1000 runs could be too long.
Let's figure out what the optimum runs / processes / text sizes are for good general measurement.
Default of 500 is fine. Longer search texts seem to be more important than number of different patterns of the same length.
Will give more benchmarking accuracy, and we have speed to spare now that algos are opened with dlopen() rather than execing a new process for each iteration.