GIL demo for Python3 using hashcash as a computational challenge

gil_demo.py

"""An example demonstrating the limitations of the GIL (CPython). This example
increments a counter for a given challenge so that the first N bits of the SHA1
hash of the challenge are all zeros (0). The computational effort increases
exponentially with the number of bits required to be zero.

This demo illustrates that this task cannot be sped up by parallelizing the
computation."""
import base64
import hashlib
import threading
import time

# send signals between threads. Used to mark when a solution was found
solution_found = threading.Event()


def are_first_n_bits_zero(byte_string: bytes, n: int) -> bool:
    """Return True if the first N bits are zero (0)"""
    byte_index = (n // 8)  # Calculate the byte index of the Nth bit
    bit_offset = 8 - (n % 8)  # Calculate the bit offset within that byte

    # Check if all the bits before the Nth bit are zero
    if any(byte_string[:byte_index]):
        return False

    # Check if the bits within the Nth byte before the Nth bit are zero
    if (byte_string[byte_index] >> bit_offset) == 0:
        return True

    return False


def sha160(challenge: bytes) -> bytes:
    """Compute the SHA1 160-bit hash"""
    return hashlib.sha1(challenge).digest()


def to_base64(v: object) -> bytes:
    """Encode any object as base64"""
    return base64.b64encode(str(v).encode())


def to_bin(digest: bytes) -> str:
    """Format a byte string as binary"""
    bits = ''
    for b in digest:
        bits += f"{b:08b}"
    return bits


def mint(challenge: str, bits: int) -> tuple[int, bytes]:
    """Mint a counter for the challenge, so that the SHA1 has begins
    with N zeros, where N=bits. Not parallelized."""
    counter = 0
    encoded_challenge = challenge.encode()
    while True:
        digest = sha160(encoded_challenge + to_base64(counter))
        if are_first_n_bits_zero(digest, bits):
            return counter, digest
        counter += 1


def worker(thread_id: int, challenge: str, bits: int, num_threads: int) -> None:
    """Parallelized version of `mint`."""
    print(f"Thread {thread_id} is starting search")
    encoded_challenge = challenge.encode()

    # Simulate some work
    for counter in range(thread_id, 2**32, num_threads):
        digest = sha160(encoded_challenge + to_base64(counter))

        if solution_found.is_set():
            print(f"Thread {thread_id} is stopping because solution was found")
            return

        if are_first_n_bits_zero(digest, bits):
            print(f"Thread {thread_id} found the solution!")
            print(f"SOLUTION: {counter}")
            solution_found.set()
            return

    print(f"Thread {thread_id} did not find the solution and is exiting")


if __name__ == '__main__':
    challenge = "Hello, my name is Leon."
    bits = 21
    num_threads = 4

    # Single threaded
    start = time.time()
    counter, digest = mint(challenge, bits)
    assert to_bin(digest)[:bits] == '0'*bits
    print('SOLUTION', counter)
    print(f'Single thread took: {time.time() - start:2f}s')

    # Create threads
    threads = [threading.Thread(target=worker, args=(i, challenge, bits, num_threads)) for i in range(num_threads)]
    start = time.time()

    # Start threads
    for t in threads:
        t.start()

    # Wait for all threads to complete
    for t in threads:
        t.join()

    print(f'Multiple thread took: {time.time() - start:2f}s')