/ptime

IPython magic for parallel profiling (like `%time`, but parallel)

Primary LanguagePythonBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

ptime

A tool for measuring serial and parallel execution, and comparing the results. Provides an IPython magic %ptime. This can be useful for measuring the benefits of parallelizing code, including measuring the effect of the Global Interpreter Lock (GIL).

Example

In [1]: %load_ext ptime

In [2]: import numpy as np

In [3]: x = np.ones((5000, 10000))

In [4]: %ptime x + x
Total serial time:   0.42 s
Total parallel time: 0.25 s
For a 1.67X speedup across 2 threads

In [5]: %ptime -n4 x + x  # use 4 threads
Total serial time:   0.82 s
Total parallel time: 0.31 s
For a 2.60X speedup across 4 threads

In [6]: res = %ptime -o x + x  # Get the result
Total serial time:   0.41 s
Total parallel time: 0.25 s
For a 1.66X speedup across 2 threads

In [7]: res.speedup
Out[7]: 1.6610825669011922

In [8]: %%ptime  # Use as a cell magic
...: x = np.ones((5000, 10000))
...: y = x + x
...:
Total serial time:   0.72 s
Total parallel time: 0.47 s
For a 1.54X speedup across 2 threads

Install

This package is available via pip:

pip install ptime