Weekly - issue 43

How to Benchmark a Python Statement

• time.perf_counter()

  from time import perf_counter
  
  # record start time
  time_start = perf_counter()
  
  # execute the statement
  data = [i*i for i in range(100000000)]
  
  # record end time
  time_end = perf_counter()
  
  # calculate the duration
  time_duration = time_end - time_start
  
  # report the duration
  print(f'Took {time_duration} seconds')

• timeit

  from timeit import timeit
  
  # benchmark the statement
  time_duration = timeit('[i*i for i in range(100000000)]', number=1)
  
  # report the duration
  print(f'Took {time_duration} seconds')

  OR

  # The command executes the statement one time and
  # reports the best time (smallest time) over 3 repetitions of executing the statement
  python -m timeit -n 1 -r 3 "[i*i for i in range(100000000)]"

How to Share Large Data Between Processes in Python

• Inherit a copy of the structure in each process

  It is the fastest way.

• Pass a copy of the structure to each task via an argument

  It is the slowest way.

• Initialize process workers with a copy of the structure once