CPU-bound thread under GIL (ARCH2042)

Key python.async.cpu-bound-thread-gil · Severity info · Category async

Spawning threads for CPU-bound work in a GIL-constrained runtime like CPython gives no speedup; threads run only one Python bytecode at a time. Use processes for CPU parallelism instead.

Incorrect

import threading

def crunch(chunks):
    threads = [threading.Thread(target=cpu_heavy, args=(c,)) for c in chunks]
    for t in threads:
        t.start()  # does not parallelize under the GIL

Correct

from concurrent.futures import ProcessPoolExecutor

def crunch(chunks):
    with ProcessPoolExecutor() as pool:
        return list(pool.map(cpu_heavy, chunks))  # real parallelism

Fix

Use ProcessPoolExecutor or multiprocessing instead of threads for CPU-bound work.