Python 3.15's JIT is now back on track (fidget-spinner.github.io)

Article summary

The Python 3.15 JIT is now back on track, having met its performance goals ahead of schedule. The JIT is about 11-12% faster on macOS AArch64 and 5-6% faster on x86_64 Linux compared to the interpreter. The success is attributed to a combination of luck, community-led efforts, and the work of core contributors. The team is now aiming to add free-threading support in 3.15 or 3.16.

Main themes

• Python JIT development
• free-threading and GIL
• performance optimization
• community involvement
• language design and complexity
• investment and funding
• concurrency and parallelism
• alternative implementations like PyPy

What commenters say

• Free-threading in Python may not be necessary or beneficial for most users, and could introduce performance penalties or crashes.
• The GIL provides a useful guarantee for Python developers, and removing it may not be worth the potential benefits.
• The lack of funding and investment in Python performance has hindered the development of a JIT compiler.
• PyPy's JIT compiler could have been adopted as the standard JIT mechanism for Python, potentially simplifying the development process.
• The complexity of the Python codebase and its dynamic features make it harder to implement a JIT compiler compared to other languages.
• The need for thread safety and concurrency in Python is not as widespread as assumed, and other solutions like subinterpreters or native code may be sufficient.
• The removal of the GIL could lead to significant performance improvements and better support for parallelism in Python.
• The development of a JIT compiler for Python is important for its performance and competitiveness with other languages.