code-timer: A Python repository from gulaki

codetimer

This is a module to measure time elapsed and count number of passes through code segments and functions.

Useful for benchmarking, testing different implementaions for performance, statistical time analysis of code, etc.

Run demo_timer.py to see usage and performance characteristics for various implementations of calculating the factorial of a number.

Timer() is a simple timer that simply measures total time and number of passes through code. Use like tick() and tock() of Matlab.

>>> t = Timer()  # starts the timer object
...
>>> t.stop()  # stops the timer object
>>> t.total  # then contains total time between creation and stop() in seconds.
>>> repr(t)  # returns time with units as string, so print(t) can be used.

A timer can be started and instructed to wait till start() method is called.

>>> t = Timer(wait=True)  # creates timer but doesn't start measuring. wait is False by default. See previous case.

>>> t.start()  # begins measuring.
...
>>> t.stop()  # stops measuring.
...
>>> t.start() # starts measuring again
...
>>> t.stop()  # stops measuring
>>> t.total  # now has total time recorded between starts and stops in seconds
>>> t.count  # contains number of times the timer was triggered.

This can be used in a loop to measure time per loop or count number of times a function was called.

>>> t = Timer(unit='s')  # to define what units to print the time in.

Options are 's' (second), 'ms' (millisecond), 'm' (minute) and 'h' (hour).

By default the unit is 'ms' but in a project the default units can be set before all usages by

>>> Timer.default_unit = 's'  # or any unit

Timer.function_timer() can be used as a decorator to measure accumulated time of a function's execution. Invoke the .timer member to get the total time and count

@Timer.function_timer
def afunction(args):
    # some process

for i in range(1000):
    afunction(args)

t = afunction.timer
print(t.total, t.count))
print('Time per loop =', t.total/t.count)

afunction.timer.reset()  # resets the timer to zero.

TimerStats.function_timer() can be used as a context manager to time a chunk of code
```
with TimerStats.function_timer as t:
    # some process
print(t)
```

TimerStats() extends Timer() by enabling simple statistical analysis on collected time data.

>>> t = TimerStats()  # initializes the timer and always waits by default. Units can be specified and default can be set like Timer().
...
>>> for i in range(10000):
...     t.start()
...     ...
...     t.stop()
>>> t.final_stats()  # call this after measurement is done to calculate mean and standard deviation
>>> t.min  # minimum lap time (s)
>>> t.max  # maximum lap time (s)
>>> t.total  # total time (s)
>>> t.count  # no. of laps
>>> t.mean  # average lap time (s)
>>> t.std  # standard deviation (s)
>>> repr(t)  # returns total time with units as string
>>> str(t)  # 'total=9604.3762 ms, min=90.9867 ms, max=100.9108 ms, mean=96.0438 ms, std=10.0969 ms.'

A call to TimerStats.stop() also returns the lap time. So if required it can be saved as lap data.

TimerStats.function_timer() is similar to function_timer decorator but enables statistical analysis on the timing data using TimerStats()

eg:

@TimerStats.function_timer
def afunction(args):
    # do something
    return something

for i in range(1000):
    retval = afunction(args)
    lap = afunction.lap  # Each lap time can be caught for further analysis

t = afunction.final_stats()  # calculates mean and std and can be accessed from the members.
print(t)  # Formatted output giving detailed information

Run test_stats.py to see statistical analysis of time data from running a function. Need Matplotlib to see the plots.

MemoryTracker() is a class to track the size of a number of objects as a function of time or iteration.

data1, data2 = [], []
mem = MemoryTracker(data1, data2)  # initialize data tracker
for _ in range(1000):
    # computations
    data1.append(stuff)
    data2.append(stuff)
    mem.getsize(data1, data2)  # get size data at this point

plt.plot(mem.times, mem.sizes)  # plot to analyze space usage

Run test_getsize.py to see some trivial analysis.

Note: Feel free to suggest features/code or report bugs if any. I dont think I have tested it enough. I hope this can become a comprehensive module for developing fast computational projects.

gulaki/code-timer

codetimer