I had heard a lot about how Python is better than PHP. Even if we discard personal preferences re: syntax and readability, the Python crowd touts its performance. I decided to run a simple test.
The task is straightforward: read input from a file and count all the occurrences of the word "abracadabra". The input is a modified 17,826-word palindrome with one of the words substituted for "abracadabra" (to make our algorithms' search less frustrating), but with every word on a new line instead of separated by a comma and a space. In order to maximize Python's performance, I added a bootstrap file which imported the original .py file containing the implementation. Upon initial execution, this would compile the original .py file into a .pyc file that contains bytecode. Even so, PHP consistently performs faster than Python.
The scripts have an optional argument: a number of iterations to run the searcher cycle over. If left blank, the searcher iterates over 100 cycles. Sample usage:
php abracadabra.php
php abracadabra.php 200
python bootstrap.py
python bootstrap.py 200
In order to collect better data, there's also runscripts.php. It will run PHP and Python versions of the algorithm for 200, 2,000, and 20,000 iterations, 100 times each, and save the results into benchmarks.txt.