[NeedForSpeed] [Python-checkins] r46083 - sandbox/trunk/stringbench sandbox/trunk/stringbench/stringbench.py

[Andrew Dalke]

> [M.-A. Lemburg]
>> Why don't you add these to pybench which is now part of Python
>> (see Tools/pybench/) ?!

Tim:
> We were wondering about that :-)  If it's OK by you, we'd like to add
> several new tests to pybench.

Feel free to add any of the tests to pybench.

The goal of stringbench is a bit different than pybench.  We wanted
a head-to-head comparison of byte string performance and unicode
string performance, as well as a way to compare different methods
which give different ways to compute the same thing

For example, here are three tests for splitting on newline

========== split 2000 newlines
723.01  871.26  83.0    "...text...".rsplit("\n") (*100)
104.61  237.68  44.0    "...text...".split("\n") (*100)
106.18  280.76  37.8    "...text...".splitlines() (*100)

This suggests that the unicode splitlines could get a
speed boost because it's almost 1/3rd the speed of the
byte implementation.  ("Suggests" because the numbers /F
gets on his new Intel box are rather different than my
oldish PPC.)

					Andrew
					dalke at dalkescientific.com