Global Interpreter Lock (GIL)
Last updated
Was this helpful?
Last updated
Was this helpful?
Was this helpful?
It's a mutex that protects access to Python objects, preventing multiple threads from executing Python bytecodes at once
CPython supports multiple threads within a single interpreter, but threads must request access to the GIL in order to access any object or function.
Necessary due to CPython's memory management is not thread-safe.
Every python object has a reference counter and when it reaches zero, the garbage collector frees the memory space.
So, in order to avoid race conditions when updating every counter, a global mutex is a legit solution.
It's controversial because it prevents multithreaded CPython programs from taking full advantage of multiprocessor systems in certain situations. So, python in practice becomes single-processed.
This shouldn't be a stopper when dealing with I/O operations. This means different threads can access the GIL while other threads wait.
However, in CPU-intensive applications, this is an issue given you can't take advantage of multiple cores.
C libraries like pandas, numpy, or numba use the CPython macros to manually release the GIL.
Another option to take advantage of multicore machines is creating processes instead of threads. However, this creates an overhead (by generating a copy of the parent memory state) and making less efficient the communication between processes (and shared variables).