There is a small performance suite which can be run with the following:
% python setup.py performance| v0.0.1 | |
|---|---|
| Python2.6 | 20.77s |
| Python2.7 | 21.13s |
| Python3.3 | 27.32s |
| Pypy | 1.333s |
The moral of the performance story is that Pypy should be used if performance is important for your use case. However, the current performance tests probably overstate Pypy's performance.