Friday 3:30 p.m.–4 p.m.

Is pypy ready for production?

Mark Rees

Audience level:
Lecture Room E4


PyPy is a Python interpreter and just-in-time compiler. In some benchmarks it has been shown to run python code faster than CPython. But what about real production code? This presentation wil discuss the results of running some of my companies python codebase under pypy.


I have followed the development of PyPy since 2004 and played with various releases to see what the PyPy team had achieved. It wasn't until the release of PyPy 1.18 that I actually ran some existing production python code under it. The perceived performance improvement of this exercise inspired me to implement a more scientific method for comparing the runtime performance of CPython and PyPy running various Python production code sets that we use in the company I work for.

At PyCon Australia 2012 I gave a presentation to discuss the results of this comparison and the positives and negatives with utilising PyPy in production. By the time of PyCon Singapore, a year would have almost passed since the original presentation and with the impending release of PyPy 2.0 final it is the perfect opportunity to revisit the benchmarks and answer the question: "Is pypy ready for production?"