The Shannon Plan: Faster CPython Initiatives for Python 3.13

In the fall of 2020, CPython core developer Mark Shannon proposed a series of performance improvements known as the "Shannon Plan" (Faster CPython), aiming to increase Python's speed fivefold within four years.

Microsoft later joined the effort, supporting developers including Mark Shannon and Guido van Rossum to work on the Faster CPython project.

Recently, Shannon and Michael Droettboom presented the roadmap for Python 3.13, highlighting three parallel work streams:

Tier 2 Optimizer : Reduce interpreter time by at least 50% through superblock generation and management.

Enable Sub‑Interpreters (PEP 554) : Build on the per‑interpreter GIL introduced in Python 3.12 to allow better parallelism without C extensions.

Memory Management Optimizations : Reduce allocation overhead and improve garbage‑collection efficiency.

The Tier 2 optimizer plan includes getting the interpreter running, generating superblocks, and implementing basic superblock management.

For sub‑interpreter support, a draft PEP 554 is being prepared and needs early approval.

Memory‑management improvements target data‑structure redesign to cut allocations and faster cyclic‑GC processing.

Regarding a JIT compiler, Shannon notes that the first step is a tracing interpreter, but a full JIT may not arrive until Python 3.13, and he remains skeptical about its necessity.

There is also internal debate about completely removing CPython's Global Interpreter Lock (GIL), with the Faster Python team positioned more favorably in that discussion.

Related links:

Faster CPython 3.13 README

Hacker News discussion