measurer 1.0.0

Easily measure Python function performance

Measurer

Easily measure Python function performance

Install

$ pip install measurer

Usage

import measure

@measure.time(repeats=100, copy=False, gc_off=True)
def a():
    pass  # Your logic here
print(a())  # 6.700000000005313e-06

def b():
    pass  # Your logic here
print(measure.time(b, repeats=100, copy=False, gc_off=True)())  # 6.199999999997874e-06

Goal

Currently, this module contains only one decorator measuring function execution time. The goal was to make performance evaluation extremely easy for non-experienced programmers, who wants to check how much resources their programs/solutions take. As a subject for future development, other types of measurers could be implemented (e.g. memory, recursive function calls etc.)

Details

Why not "timeit" ?

Timeit is not suitable for function calls with arguments. Thus, one has to wrap a tested function with another function (without arguments) and evaluate redundant function calls (which sometimes causes big overhead).

We still perform redundant calls...

Yes. Ideally, we have to analyze syntactic tree and create some sort of "inline" method call. However, this is a hobby module, so we don't push towards ideal implementation.

Why do we dump parameters with Pickle?

Packages like "msgpack" or "marshal" show comparable performance (and time to time they are faster), but the difference is not significant. So we use pickle, cause it's a built-in package (we try to leave the module simple).