Results

# Tested over /mnt/redhat/rel-eng/OpenShiftEnterprise/4.4/images/cdn
# with realistic environment
result-both-sums.txt:real	0m13.294s
result-both-sums.txt:real	0m15.049s
result-both-sums.txt:real	0m14.133s
result-both-sums.txt:real	0m13.102s
result-both-sums.txt:real	0m14.525s
result-both-sums.txt:real	0m13.979s
result-both-sums.txt:real	0m15.156s
result-both-sums.txt:real	0m15.098s
result-both-sums.txt:real	0m13.009s
result-md5-only.txt:real	0m5.803s
result-md5-only.txt:real	0m5.308s
result-md5-only.txt:real	0m4.320s
result-md5-only.txt:real	0m6.257s
result-md5-only.txt:real	0m4.167s
result-md5-only.txt:real	0m4.937s
result-md5-only.txt:real	0m4.974s
result-md5-only.txt:real	0m4.269s
result-md5-only.txt:real	0m4.876s
result-no-sums.txt:real	0m0.789s
result-no-sums.txt:real	0m0.787s
result-no-sums.txt:real	0m0.808s
result-no-sums.txt:real	0m0.831s
result-no-sums.txt:real	0m0.803s
result-no-sums.txt:real	0m0.781s
result-no-sums.txt:real	0m0.741s
result-no-sums.txt:real	0m0.736s
result-no-sums.txt:real	0m1.206s
result-sha256-only.txt:real	0m9.606s
result-sha256-only.txt:real	0m10.604s
result-sha256-only.txt:real	0m10.876s
result-sha256-only.txt:real	0m10.786s
result-sha256-only.txt:real	0m11.937s
result-sha256-only.txt:real	0m10.118s
result-sha256-only.txt:real	0m10.904s
result-sha256-only.txt:real	0m10.816s
result-sha256-only.txt:real	0m10.892s

sumperf

#!/usr/bin/env python
# Testing performance of checksum calculation.
# Usage:
#
#  ./sumperf [<sumtype1> <sumtype2> ...] [<filename1> <filename2> ...]
#
# - Omit any sumtypes to test read without checksum.
# - Make sure to run it several times and/or be aware of cache
import logging
import hashlib
import sys

logging.basicConfig(level=logging.INFO, format='%(asctime)s %(message)s')

CHUNKSIZE = 1024*1024*16
LOG = logging.getLogger()
sumtypes = []
filenames = []

for arg in sys.argv[1:]:
  try:
    # test if it's a checksum type
    hashlib.new(arg)
    # OK, it is
    sumtypes.append(arg)
  except ValueError:
    # it's not, then it ought to be a filename
    filenames.append(arg)


for filename in filenames:
  count = 0
  hashers = [(sumtype, hashlib.new(sumtype)) for sumtype in sumtypes]

  LOG.info("Begin read %s", filename)

  with open(filename, 'rb') as f:
    while True:
      chunk = f.read(CHUNKSIZE)
      if not chunk:
        break
      count += len(chunk)
      for h in hashers:
        h[1].update(chunk)

  LOG.info("End read %s, bytes: %s", filename, count)
  for h in hashers:
    LOG.info("  %s: %s", h[0], h[1].hexdigest())
  LOG.info(" ============================ \n")