#!/usr/bin/env python3

import argparse
from concurrent.futures import ThreadPoolExecutor
import time

parser = argparse.ArgumentParser()
parser.add_argument('--size', '-s', type=int, default=8192, help='bytes in each chunk')
parser.add_argument('--chunks', '-n', type=int, default=8192, help='number of chunks to join')
parser.add_argument('--threads', '-t', type=int, default=1, help='number of threads')
parser.add_argument('--repeats', '-r', type=int, help='number of times to join')
args = parser.parse_args()
if args.repeats is None:
    args.repeats = max(3, 10**9 // (args.size * args.chunks))

def one_thread(args):
    chunks = [b'x' * args.size for _ in range(args.chunks)]
    start = time.monotonic()
    for repeat in range(args.repeats):
        b''.join(chunks)
    elapsed = time.monotonic() - start
    return elapsed

pool = ThreadPoolExecutor(args.threads)
futures = [pool.submit(one_thread, args) for _ in range(args.threads)]
elapsed = 0.0
for future in futures:
    elapsed += future.result()
avg_elapsed = elapsed / args.threads
rate = args.size * args.chunks * args.repeats / avg_elapsed
rate_mb = rate / 1e6
print(f'{args.size},{args.chunks},{args.threads},{rate_mb:.2f}')