The initiative to release GIL from Python, enabling full multithreaded execution on every core of the CPU, is progressing following the integration of the code into the Python project in the past March. Currently, the team from Quansight Labs has unveiled the binary of Python version Free Threading, which now fully utilizes every core of the CPU.
Python code that releases GIL must be compiled with the –disable-gil option in order to utilize this feature. The Free Threading project will constantly enable this feature regardless of whether the main Python project decides to implement this feature in Python 3.13 or not.
It is expected that the process of releasing GIL will take several more years before various modules fully support it and can truly remove the GIL from the code. Quansight has encountered issues while testing large modules such as numpy or pywavelets. Additionally, reports of problems have also been found in modules like scipy and Pillow.
Downloads for testing are available on different platforms including Windows, macOS, and various Linux distributions.
Source – Quansight
TLDR: Quansight Labs is making progress in releasing GIL from Python to enable full multithreaded execution on all CPU cores with the introduction of Python version Free Threading. Testing and porting modules to support this feature is expected to take several more years before GIL can be completely removed from the code. Various large modules like numpy and pywavelets have encountered issues during testing, and reports of problems have also been identified in modules like scipy and Pillow. Test downloads are available for multiple platforms including Windows, macOS, and various Linux distributions.
Leave a Comment