Global interpreter lock

Category: Parallelism · Status: stub · Lessons: chapter 07, Parallelism

One line: A lock that lets only one thread run interpreter code at a time — CPython's, and Ruby's — so threads give concurrency but not CPU parallelism; CPython 3.13 added an optional build without it.

Also called: GIL, GVL, free threading.

How it connects

flowchart LR
  n_gil["Global interpreter lock"]
  n_mutex["Mutex"]
  n_gil -->|uses| n_mutex
  classDef center stroke-width:3px
  class n_gil center
  classDef outside stroke-dasharray: 4 3
  class n_mutex outside

• Is built on: Mutex
• See also: I/O-bound and CPU-bound work, Multiprocessing

In each language

Python	CPython's GIL ↗; the free-threaded build ↗, available since 3.13, disables it
Elsewhere	CRuby holds its Global VM Lock (GVL) per Ractor ↗, so ractors run in parallel where threads cannot

Where to read more

• In the books: Python Concurrency with asyncio, Matthew Fowler — ch. 1, 'Getting to know asyncio' → 'Understanding the global interpreter lock'
• In the books: Parallel Programming with Python, Jan Palach — ch. 1, 'Contextualizing Parallel, Concurrent, and Distributed Programming' → 'Taking care of Python GIL'
• In the books: Advanced Python Programming, Quan Nguyen — ch. 15, 'The Global Interpreter Lock'
• In the books: Fluent Python, Luciano Ramalho — ch. 19, 'Concurrency Models in Python' → 'The Real Impact of the GIL'
• In the books: Python Cookbook, David Beazley, Brian K. Jones — ch. 12, 'Concurrency' → 'Dealing with the GIL (and How to Stop Worrying About It)'
• Reference: PEP 703: Making the Global Interpreter Lock Optional in CPython ↗