stt.py

"""
Time test code from Sheridan Green's Python Multiprocessing jupyter notebook
https://github.com/galastrostats/general/blob/master/PythonMultiprocessing.ipynb

Minimally modified by Sheila Kannappan June 2017 to allow the user to enter 
%timeit stt.main(nproc)
on the ipython command line, where nproc is between 1 and N-1 for an N-core machine
"""

import multiprocessing
import time

def runSimulation(params):
    """This is the main processing function. It will contain whatever
    code should be run on multiple processors.
    
    """
    param1, param2 = params
    # Example computation
    processedData = []
    for ctr in range(100000):
        processedData.append(param1 * ctr - param2 ** 2)

    return processedData

def main(nproc):
    # Define the parameters to test
    param1 = range(100)
    param2 = range(2, 202, 2)

    params = zip(param1, param2)

    pool = multiprocessing.Pool(nproc) # number of processes

    # Parallel map
    tic = time.clock()
    results = pool.map(runSimulation, params)
    toc = time.clock()

    # Serial map
    tic2 = time.clock()
    results = map(runSimulation, params)
    toc2 = time.clock()

    print('Parallel processing time: %r\nSerial processing time: %r'
          % (toc - tic, toc2 - tic2))
    
    print('Speedup factor: %r') % ((toc2-tic2)/(toc-tic))
    
if __name__ == '__main__':
    main()