Numba is a Python JIT compiler which compiles Python bytecodes to binary code and can run the code on the CPU in parallel
and optionally also on the GPU (Cuda).
Additionally it allows to code Universal Functions for Numpy arrays in python (they will also be JIT-compiled to binary code).
With that you can get pretty close to VEX speed in some conditions and have a bit more “Python-freedom” at the same time
…Nothing beats VEX of course…
The first Video is about using it & its features in general & on Linux.
The second Video is about installing it on Windows.
1. Video link: How to use Numba (Python JIT) with Houdini (General & Linux) [vimeo.com]
2. Video link: How to use Numba (Python JIT) with Houdini (Windows) [vimeo.com]
(I have catched a bad flue which killed my voice, so I had to use only annotations in the videos…
)I have provided the “FastPython” node plugin in my other forum thread [www.sidefx.com] for Linux & Windows.
Let me know if you are interested in the recompiled Python modules needed for using Numba with Houdini on Windows.
I think I could make them available for download somewhere.