Which Linux Deletion Method Is Fastest for Half a Million Files?

We created 500,000 tiny text files and measured how quickly each tool could delete them on a Linux machine.

1. rm

$ time rm -f *

The command fails with "argument list too long"; the reported timing is 3.63 s user, 0.29 s system, 98 % CPU, total 3.985 s, but it cannot handle the massive file count.

2. find with -exec rm

$ time find ./ -type f -exec rm {} \;

This method took roughly 43 minutes (49.86 s user, 1032.13 s system, 41 % CPU, total 43:19.17).

3. find with -delete

$ time find ./ -type f -delete

Completed in about 9 minutes (0.43 s user, 11.21 s system, 2 % CPU, total 9:13.38).

4. rsync --delete

# First create an empty directory blanktest
$ time rsync -a --delete blanktest/ test/

Finished in roughly 16 seconds (0.59 s user, 7.86 s system, 51 % CPU, total 16.418 s), making it both fast and convenient.

5. Python script

import os, timeit

def main():
    for pathname, dirnames, filenames in os.walk('/home/username/test'):
        for filename in filenames:
            file = os.path.join(pathname, filename)
            os.remove(file)

if __name__ == '__main__':
    t = timeit.Timer('main()', 'from __main__ import main')
    print(t.timeit(1))

The script ran for about 9 minutes (529.3 seconds).

6. Perl one‑liner

$ time perl -e 'for(<*>) { ((stat)[9] < (unlink)) }'

Completed in roughly 16 seconds (1.28 s user, 7.23 s system, 50 % CPU, total 16.784 s), the fastest among the scripting approaches.

Result Summary

rm: not usable for huge file count
find -exec: 43 minutes
find -delete: 9 minutes
Perl: 16 seconds
Python: 9 minutes
rsync --delete: 16 seconds

Conclusion: Using rsync with the --delete option is the quickest and most convenient way to delete a massive number of small files on Linux.