Posts tagged numba

Benchmarking Python Flavors: PyPy, CPython (standard / free-threaded / jit), Cython & Numba

Now that Python 3.14 is out, let’s compare how CPython fares against PyPy! Python has many implementations and optimization tools. This repository contains a focused benchmarking suite that compares several of them across three workload classes — compute-bound, I/O-bound, and parallel workloads. The goal of this article is to explain what was measured, how the experiments are organized, and what the results typically show when you run the suite.

Read more ...


Exploring Python Performance: PyPy vs CPython vs Numba

Python developers often face the decision of choosing between different implementations for their projects, especially when performance is a crucial factor. In this blog post, we delve into a comparative benchmarking study between PyPy, CPython, and Numba, focusing on calculating Pi value using a custom script.

Read more ...


Calculating π with Numba

Here is a version of the calculation of π using numba. Numba does just-in-time compilation from Python code and also supports parallelism. I was able to test calculate_pi() in the order of billions of iteretions and without consuming too much memory.

Read more ...