Explain the GIL and how you would speed up a CPU-bound Python job.
Answer
- GIL: a lock in CPython; only one thread executes Python code at a time. It protects reference counts and interpreter state.
- It is released during blocking I/O and by many C extensions (NumPy, hashlib, zlib).
- I/O-bound work: threads or
asyncio work well. - CPU-bound options:
ProcessPoolExecutor / multiprocessing; vectorize with NumPy/pandas; move the hot loop to Cython/Rust/C; free-threaded build (3.13+) where libraries support it. - Trade-offs: process start-up, pickling cost, memory duplication. Profile first (
cProfile, py-spy).
Practice this question in Repaso
