gtimeit

Graphical code timer, and function time usage tracker. A tool for python functions. Time and compare your functions inside or outside a program.

Installation

Installing as library

The easiest way is to use pip.

# install the library with plotting capabilities (requires matplotlib)
pip install gtimeit[plot]
# or without
pip install gtimeit

You can now import the tools in your code.

from gtimeit import Benchmark
from gtimeit import execTimeList, display_performances

Installing for developpement

Or you can install from Github.

git clone https://github.com/Chthi/gtimeit
cd gtimeit
pip install --editable ".[plot,dev]"

Benchmark

Allows to run multiple python functions a certain amount of times and to compare there respective execution times. Similar to timeit Display curves of all the execution times. Prototype : benchmarking for complexity tests up to one variable parameter. The variable take all the values of a given range and the benchmarck keeps track of the executions times.

Usage :

bm = Benchmark()
bm.add(fct1, *args, **kwargs)
bm.add(fct2, *args, **kwargs)
bm.add(fctn, *args, **kwargs)
bm.run(20, multiplicator=100)

Quick usage :

bm = Benchmark([fct1, fct2], 300)

Examples

More examples available in benchmark_examples.py.

def plusEgale(i=1):
    i += 1
    return i

def plusPuisEgale(i=1):
    i = i + 1
    return i

# running 200 tests 100000 times each
bm = Benchmark()
bm.add(plusEgale, 42)
bm.add(plusPuisEgale, 42)
bm.run(200, multiplicator=100000)

# short version for procedures
bm = Benchmark([plusEgale, plusPuisEgale], 200, multiplicator=100000)

Tracker

Allow to keep track of the execution time of some functions in a program. Useful to know which part of your program take the more time to execute.

Usage

Use the decorator @execTimeList before every function you want to keep track of. The execution time of a function is defined by the total execution time of the code and of the functions called inside that are not taged with @execTimeList. Use display_performances() to display the results.

Examples

from gtimeit import execTimeList, display_performances

@execTimeList
def additions():
    i = 0
    for x in range(1,100000):
        i += 1

@execTimeList
def sumrange():
    sum(range(1,100000))

@execTimeList
def extending():
    l = []
    for x in range(1,100000):
        l.extend([0])

@execTimeList
def fct1():
    extending()
    sumrange()

# simulate some basic code
for x in range(4):
    additions()
    sumrange()
    extending()
for x in range(2):
    sumrange()
    extending()
for x in range(10):
    additions()
fct1()

display_performances()

Limitations

For the current version :

• It is not possible to use Benchmark and execTimeList decorator at the same time on one object as the decorator replace the function name by timed.
• Recursive tracked functions are not supported. If a function decorated with @execTimeList calls itself (directly or through mutual recursion) on the same thread, its exclusive time is misattributed: the accumulated durations all collapse onto the innermost call while the outer frames are left at zero. The number of calls, wall_duration and total time stay correct; only the exclusive-time views (pie chart, timeline) are affected.