Scribd
Upload
3 views7 pages

11 Benchmarking in Python

The document discusses the importance of benchmarking algorithms in Python, emphasizing the need for empirical performance comparisons due to various influencing factors. It explains how time is represented in Python and provides methods for benchmarking both single and multiple runs. Additionally, it includes useful references for further reading on Python's time documentation and benchmarking techniques.

Uploaded by

Valery Taustsiakou
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
3 views7 pages

11 Benchmarking in Python

The document discusses the importance of benchmarking algorithms in Python, emphasizing the need for empirical performance comparisons due to various influencing factors. It explains how time is represented in Python and provides methods for benchmarking both single and multiple runs. Additionally, it includes useful references for further reading on Python's time documentation and benchmarking techniques.

Uploaded by

Valery Taustsiakou
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Benchmarking Algorithms in Python

Dr Patrick Mannion
[email protected]

Dr Patrick Mannion, Dept of Computer Science & Applied Physics 1


Overview
• Motivation for benchmarking
• Time in Python
• Benchmarking a single run
• Benchmarking multiple statistical runs

Dr Patrick Mannion, Dept of Computer Science & Applied Physics 2


Motivation for benchmarking
• Benchmarking or a posteriori analysis is an empirical method to
compare the relative performance of algorithm implementations
• Experimental (e.g. running time) data may be used to validate
theoretical or a priori analysis of algorithms
• Various hardware and software factors such system architecture, CPU
design, choice of Operating System, background processes, energy
saving and performance enhancing technologies etc. can affect
running time
• Therefore it is prudent to conduct multiple statistical runs using the
same experimental setup, to ensure that your set of benchmarks are
representative of the performance expected by an “average” user
Dr Patrick Mannion, Dept of Computer Science & Applied Physics 3
Time in Python
• Dates and times in Python are represented as the number of seconds
that have elapsed since midnight on January 1st 1970 (the “Unix
Epoch”)
• Each second since the Unix Epoch has a specific timestamp
• Can import the time module in Python to work with dates and times
• e.g. start_time = time.time() # gets the current time in seconds
• e.g. 1555001605 is Thursday, 11 April 2019 16:53:25 in GMT

Dr Patrick Mannion, Dept of Computer Science & Applied Physics 4


Benchmarking a single run

Dr Patrick Mannion, Dept of Computer Science & Applied Physics 5


Benchmarking multiple statistical runs

Dr Patrick Mannion, Dept of Computer Science & Applied Physics 6


Useful references
• Python 3 time documentation:
https://docs.python.org/3/library/time.html
• Python Date and Time overview:
https://www.tutorialspoint.com/python/python_date_time.htm

• Discussions of benchmarking issues in Python (advanced material):


https://www.peterbe.com/plog/how-to-do-performance-micro-
benchmarks-in-python
https://www.blog.pythonlibrary.org/2016/05/24/python-101-an-
intro-to-benchmarking-your-code/
Dr Patrick Mannion, Dept of Computer Science & Applied Physics 7

You might also like