Program to benchmark set of checksum commands

sumbench

#!/usr/bin/python3
'Program to benchmark set of checksum commands.'
# Author: M.Blakeney, Jul 2018.

import os, argparse, time, tempfile, subprocess, statistics
import collections, platform

# Default set of commands to execute
CMDS = 'md5sum,sha1sum,sha256sum'

# Default functions to use for time calc and average
TIMEFUNC = time.perf_counter
AVEFUNC = statistics.median

# Process command line options
opt = argparse.ArgumentParser(description=__doc__.strip())
opt.add_argument('-c', '--count', type=int, default=5,
        help='number of times to run each command, default=%(default)d')
opt.add_argument('-m', '--mb', type=int, default=100,
        help='MB size of file to use, default=%(default)d')
opt.add_argument('--cmds', default=CMDS,
        help='set of commands to run, default="%(default)s"')

args = opt.parse_args()

# Create temp file with random content of required size
fp = tempfile.NamedTemporaryFile()
fp.write(os.urandom(args.mb * 1024 * 1024))
fp.flush()

cmds = [cmd.strip() for cmd in args.cmds.split(',')]
results = collections.defaultdict(list)

# Do the sample runs
for count in range(args.count):
    for cmd in cmds:
        tstart = TIMEFUNC()
        subprocess.run([cmd, fp.name], stdout=subprocess.DEVNULL)
        results[cmd].append(TIMEFUNC() - tstart)

# Calculate a representative time from lists of sample run results
times = {k: AVEFUNC(v) for k, v in results.items()}
stimes = sorted(times, key=times.get)
base = times[stimes[0]]
host = platform.platform().strip()

print(f'Results for {args.count} runs for {args.mb}MB file on {host}:')
for ind, cmd in enumerate(stimes, 1):
    val = times[cmd]
    percent = (val - base) * 100 / base
    print(f'{ind:2}: {val:8.3f}s {percent:+8.1f}% {cmd}')