[SciPy-Dev] Numba as a dependency for SciPy?

[Serge Guelton]

> Overall I'd say that:
> - Numba is better than Cython at: performance, ease of use, size of generated
> binaries, and templating support for multiple dtypes. Possibly also maintenance
> status right now.

Hi all,

To assert that statement, I've quickly setup a bunch of notebooks with
cython kernels extracted from the scipy codebase here:

    https://github.com/serge-sans-paille/scipy-kernels

For the sake of the comparison, I've contributed Pythran implementation
of these kernels, and I hope someone in the ML will issue a PR with
numba version.

Please note that the goal is *not* to claim that a given compiler is
better than another. The benchmarks provided only target a subset of the
functions input space, I have not tuned the backend compiler for a given
architecture nor tried to enable parallelism.

It's just there to give a rough idea of the ease-of-use / performance
tradeoffs.

Best,

Serge