Control number of iterations

[pelegm]

How many iterations are done by default in benchmarking? I ask this because adding @bench(...) to one of my tests has increased its total duration from about 1.34s to about 120s, but the mean running time was only 3.6us...

Is it possible to add a command-line argument, or a decorator argument, that will control the number of iterations done for benchmarking?

[mehcode]

You should be able to do:

@pytest.mark.bench('method', iterations=10)

[pelegm]

Thanks! For those who have not looked into the code, the default number of iterations is 100.

So: one questions is left: how is it that even when I put iterations=1, I get the line

tests/test_real.py: test_real_binomial ... 1.91           3.10           2.86       1.08

in which the min and the mean are different?