An overview of the compiler and runtime pipeline that Numba uses to take Python source code, turn it into a CUDA kernel, and launch it. It pulls in a variety of Numba internals to illustrate how the different parts of the pipeline work.
Contents:
- Life of a Numba Kernel: The notebook ready to execute.
- Life of a Numba Kernel with output: The notebook with output - good for viewing without needing to compile / execute, perhaps if you don't have a CUDA device or installation of Numba handy.
- life-of-a-numba-kernel.py: Example code executable as a Python script.