Deleting Files

Delete huge number of files in minutes/hours

find /path/to/folder/having/all/files -mindepth 1 -maxdepth 1 -print0 | parallel -0 rm -rvf

Example: find ./seq/ -mindepth 1 -maxdepth 1 -print0 | parallel -0 rm -rvf

Command Explanation:

• find ./seq/ => combining the find command with GNU Parallel to run rm (remove) in parallel on each immediate subfolder or file inside ./seq

• -mindepth 1 => skip the top-level directory (here, ./seq/), and only return things inside it. So, if ./seq/ contains folder1, folder2, etc., those are the targets.

• -maxdepth 1 => only look at immediate children (1 level deep), not recursively. So you get only (./seq/folder1, ./seq/folder2, ./seq/file1.txt)… not files inside those folders.

• -print0 => outputs each result terminated with a null character (\0) instead of a newline. Because filenames can contain spaces, newlines, or other weird characters. Using -print0 makes sure they're passed safely to the next command.

• | parallel -0 => puts the null-separated results to GNU parallel (which uses multiple CPU cores). The -0 tells parallel to expect input to be null-separated (to match -print0). This avoids filename issues, same as above.

• rm -rvf => this is the command parallel runs on each item it receives.

  • rm: remove files/directories.

  • -r: recursive – delete folders and everything inside.

  • -v: verbose – print what’s being deleted (this slows down the command a little bit).

  • -f: force – ignore errors, don't prompt for confirmation.

So for each item, parallel runs something like:

rm -rvf ./seq/folder1
rm -rvf ./seq/file2

…simultaneously across multiple CPU cores.

Even More Speed

• Drop the -v flag:

find ./seq/ -mindepth 1 -maxdepth 1 -print0 | parallel -0 rm -rf

• Use --jobs N to control how many processes run at once (Example, use 8 cores):

... | parallel -0 -j 8 rm -rf

Make parallel to utilize all the available cores/thread for max speed

• GNU parallel is smart enough to auto-detect the number of CPU cores

find ./seq/ -mindepth 1 -maxdepth 1 -print0 | parallel -0 rm -rf

• Use as many jobs as your system can handle (nproc cores, by default).

• Avoid output clutter (no -v).

• Delete in blazing-fast parallel, safely.

• Use total number of CPU in your system directly

use nproc command to find total number of CPU in your system

Squeeze every ounce of Speed

find ./seq/ -mindepth 1 -maxdepth 1 -print0 | parallel -0 -j "$(nproc)" rm -rf

Or hardcode after executing nproc in terminal:

... | parallel -0 -j 96 rm -rf