Welcome to PyPy's fork of Numpy, NumPyPy. In order to install, first install PyPy, hints are here Note this is a binary install, no lengthy translation or compilation necessary. Once you have pypy working and feel comfortable using it, you can install our version of the numpy module into a virtual environment in a separate directory:

virtualenv -p /path/to/pypy/bin/pypy /directory/to/try/pypy-numpy
git clone;
cd numpy; /director/to/try/pypy-numpy/bin/pypy install

or without a git checkout:

virtualenv -p /path/to/pypy/bin/pypy /directory/to/try/pypy-numpy
 /directory/to/try/pypy-numpy/bin/pip install git+

If you are using PyPy 4.0.1 (and not a nightly build), you must checkout the pypy-4.0.1 tagged revision, so replace the git line above with:

git clone; git checkout pypy-4.0.1

If you get a message about missing Python.h you must install the pypy-dev package for your system

If you get a message about "unable to find vcvarsall.bat", you need to install install a compiler. Microsoft has a download for that at

If you installed to a system directory, you may need to run:

sudo pypy -c 'import numpy'

once to initialize the cffi cached shared objects as root

For now, NumPyPy does not work with PyPy3*, and is not complete. You may get warnings or NotImplemented errors. Please let us know if you get crashes or wrong results.

If you do not have lapack/blas runtimes, it may take over 10 minutes to install, since it needs to build a lapack compatability library. However, you may later install upstream compatible runtimes, and NumPyPy should pick them up automatically the next time you run PyPy.

Also note that the latest version of NumPyPy will probably not run in an older PyPy. Specifically, we require cffi 1.0 or later. Since cffi is baked into PyPy, you cannot update cffi in any version of PyPy (true as of Nov 2015) so there is no recourse but to update PyPy.

The original README.txt follows:

NumPy is the fundamental package needed for scientific computing with Python. This package contains:

  • a powerful N-dimensional array object
  • sophisticated (broadcasting) functions
  • tools for integrating C/C++ and Fortran code
  • useful linear algebra, Fourier transform, and random number capabilities.

It derives from the old Numeric code base and can be used as a replacement for Numeric. It also adds the features introduced by numarray and can be used to replace numarray.

More information can be found at the website:

After installation, tests can be run with:

python -c 'import numpy; numpy.test()'

The most current development version is always available from our git repository: