Why is my Python multiprocessing code slower than single-thread execution?

Forking off four cPython interpreters, and doing import multiprocessing in each, will take non-zero time. This hardly seems like a fair comparison with the single-threaded case, which the OP does not present.

the multiprocessing version runs Slower than the single-threaded version.

That hardly seems surprising, given that you asked it to do more than the single-threaded version. Plus, you did not describe the benchmark environment, nor the elapsed times.

Starting up a new cPython interpreter does not happen for free. There's a lot of initialization work to attend to. You have posed an XY question. Perhaps in future you will choose to tell us what your true goal might be.

This can happen when the time needed to construct/execute/destroy the subprocess(es) is longer than the time needed to execute the compute() function.

However, if we extend your code to include a single-threaded approach, the multiprocessing variant is, in fact, faster.

from multiprocessing import Pool import time N = 10_000_000 R = 4 def compute(n): total = 0 for i in range(n): total += i*i return total def mproc(): start = time.perf_counter() with Pool(R) as p: p.map(compute, [N]*R) return time.perf_counter() - start def tproc(): start = time.perf_counter() for _ in range(R): compute(N) return time.perf_counter() - start if __name__ == "__main__": for func in (mproc, tproc): d = func() print(func.__name__, f"{d:.4f}s")

Output:

mproc 0.4869s

tproc 1.3456s

Python 3.14.2 on Apple Silicon (M2) MacOS 26.2