Skip to content

Latest commit

 

History

History
125 lines (90 loc) · 3.48 KB

README.md

File metadata and controls

125 lines (90 loc) · 3.48 KB
Raw

timer

A simple timer for profiling python code.

Installation

pip install git+https://github.com/camall3n/timer.git

Usage

Timers can be used in a few different ways. The simplest of these is to wrap the computation block you'd like to time in a with statement:

from timer import Timer

with Timer('my_func'):
    func()

Alternatively, you can use @Timer.wrap() to decorate a function:

@Timer.wrap()
def my_func(*args):
    # do stuff
    return

@Timer.wrap('foo')
def foo_func(*args):
    # do stuff
    return

The decorator syntax automatically uses the function's qualified name as a tag (e.g. __classname__.__name__), but it accepts an optional tag argument if you'd like to override this with your own identifier (e.g. 'foo'). Decorator syntax is equivalent to wrapping all occurrences of a function call using the with syntax and specifying the same tag for each occurrence.

Note the use of parentheses in @Timer.wrap(). Normal decorators don't need parentheses, but Timer.wrap() is actually a decorator-generating function, which is why it needs them.

Printing Statistics

The timer module will attempt to automatically print statistics at the end of program execution. To print manually, use Timer.stats().

----------------------------------------------------------
tag          frac       time    percall      rate    calls
-------  --------  ---------  ---------  --------  -------
my_func  0.857573  1.000504s  1.000504s  0.999496        1
----------------------------------------------------------
Total time: 1.166669s
----------------------------------------------------------

This table displays (left to right):

  • tag: the identifier string for the computation block or function
  • frac: the fraction of the total time that the computation block takes up
  • time: the corresponding wallclock time
  • percall: the average time per call
  • rate: the effective rate of calls per second
  • calls: the total number of calls

The stats are global, so you can import timer wherever you need it and the stats will print for all timers. The total time since the timer module was first imported is displayed at the bottom of the table.

To delete all timers and reset all statistics to zero, use Timer.reset().

Example

The following example is available in the test/ directory.

Command:

python test/test.py

Source:

import time
from timer import Timer

N = 1000

def a(i):
    with Timer('a'):
        return sum([x for x in range(i)])

@Timer.wrap('my_func')
def b():
    time.sleep(0.01)

class Thing:
    @Timer.wrap()
    def do_stuff(self):
        time.sleep(.3)

def c():
    time.sleep(1)

def main():
    [b() for i in range(N) if a(i) < N]

    Thing().do_stuff()

    with Timer('c'):
        c()

if __name__=='__main__':
    main()

Output:

---------------------------------------------------------------------
tag                 frac       time    percall          rate    calls
--------------  --------  ---------  ---------  ------------  -------
a               0.026705  0.051122s  0.000051s  19561.160340     1000
my_func         0.285803  0.547121s  0.011894s     84.076532       46
Thing.do_stuff  0.158690  0.303786s  0.303786s      3.291793        1
c               0.525156  1.005321s  1.005321s      0.994707        1
---------------------------------------------------------------------
Total time: 1.914330s
---------------------------------------------------------------------