Profiling Python code – as a whole or by function – shows where you should spend time speeding up your programs.

Python finally has interoperable tools for programming the GPU – with or without CUDA.

Although Python is a popular language, in the high-performance world, it is not known for being fast. A number of tactics have been employed to make Python faster. We look at three: Numba, Cython, and ctypes.

In this third and last article on OpenMP, we look at good OpenMP coding habits and present a short introduction to employing OpenMP with GPUs.