Overview
Explore a groundbreaking approach to achieving true parallelism in Python through subinterpreters and immutability in this 30-minute PyCon US talk. Discover the innovative "memhive" framework, which implements a worker pool of subinterpreters, an efficient RPC mechanism, and async/await-ready API. Learn about fundamental data structures like tuples, mappings, and scalar types that enable efficient structured data sharing between Python subinterpreters. Gain insights into unlocking parallelism without the overhead of data serialization, and understand the underlying algorithms that make this possible. Access the presentation slides for a comprehensive overview of this novel technique to overcome the Global Interpreter Lock (GIL) in Python.
Syllabus
Talks - Yury Selivanov: Overcoming GIL with subinterpreters and immutability
Taught by
PyCon US