Home

Python's multi-core story is murky at best. Not only can we be more clear on the matter, we can improve Python's support. The result of any effort must make multi-core (and concurrency) support in Python obvious, unmistakable, and undeniable (and keep it Pythonic).

The goal of this project is to do just that, by using the existing subinterpreter C-API. The minimal multi-core solution will involve:

• resolving existing bugs and blockers
• exposing the existing support (from C-API) in a stdlib module
• adding a mechanism to safely "share" basic immutable objects between interpreters
• no longer sharing the GIL between interpreters

At a high level, we're doing the following concrete tasks:

• PEP 554 (Multiple Interpreters in the Stdlib)
• improve interpreter isolation (move fields from PyRuntimeState to PyInterpreterState, including the GIL)

Past Discussion

• (2012) Nick Coghlan blog: Volunteer developed free-threaded cross platform virtual machines?
• (2015) Eric Snow on python-ideas: solving multi-core Python
• (2016) Eric Snow blog: Solving Multi-Core Python
• (2017) Eric Snow on python-ideas: Exposing CPython's subinterpreter C-API in the stdlib.
• (2018) Eric Snow at language summit: Subinterpreter support for Python

Misc