Use processes (not threads) for CPU-bound work to sidestep the GIL. Functions must be picklable — define them at module level, not as locals or lambdas.
Python
from concurrent.futures import ProcessPoolExecutor
import hashlib
def heavy(n: int) -> str:
"""Pure-CPU work — many threads would serialize on the GIL."""
h = hashlib.sha256(b"seed").digest()
for _ in range(n):
h = hashlib.sha256(h).digest()
return h.hex()[:12]
if __name__ == "__main__":
inputs = [200_000] * 8
with ProcessPoolExecutor() as pool:
for digest in pool.map(heavy, inputs):
print(digest)
# On a 4-core machine: ~4x faster than threads for this kind of workSave your own code snippets
Create a free account and build your private vault. Share publicly whenever you want.