GNU parallel
GNU parallel is a command-line driven utility for Linux and other Unix-like operating systems which allows the user to execute shell scripts or commands in parallel. GNU parallel is free software, written by Ole Tange in Perl. It is available under the terms of GPLv3.[2]
Developer(s) | GNU Project |
---|---|
Stable release | 20191122[1]
/ 22 November 2019 |
Repository | |
Written in | Perl |
Operating system | GNU |
Type | Utility |
License | GPLv3 |
Website | www |
Usage
The most common usage is to replace the shell loop, for example
while read x; do
do_something "$x"
done < list | process_output
to the form of
< list parallel do_something | process_output
where the file list
contains arguments for do_something
and where process_output
may be empty.
Scripts using parallel are often easier to read than scripts using pexec.
The program parallel features also
- grouping of standard output and standard error so the output of the parallel running jobs do not run together;
- retaining the order of output to remain the same order as input;
- dealing nicely with filenames containing special characters such as space, single quote, double quote, ampersand, and UTF-8 encoded characters;
By default, parallel runs as many jobs in parallel as there are CPU cores.
Examples
find . -name "*.foo" | parallel grep bar
The above is the parallel equivalent to:
find . -name "*.foo" -exec grep bar {} +
This searches in all files in the current directory and its subdirectories whose name end in .foo
for occurrences of the string bar
. The parallel command will work as expected unless a file name contains a newline. In order to avoid this limitation one may use:
find . -name "*.foo" -print0 | parallel -0 grep bar
The above command uses the null character to delimit file names.
find . -name "*.foo" | parallel -X mv {} /tmp/trash
The above command expands {}
with as many arguments as the command line length permits, distributing them evenly among parallel jobs if required. This can lower process overhead for short-lived commands that take less time to finish than they do to launch.
find . -maxdepth 1 -type f -name "*.ogg" | parallel -X -r cp -v -p {} /home/media
The command above does the same as:
cp -v -p *.ogg /home/media
However, the former command which uses find
/parallel
/cp
is more resource efficient and will not halt with an error if the expansion of *.ogg is too large for the shell.
Wikimedia Commons has media related to GNU parallel. |
References
- Tange, Ole (22 Nov 2019). "GNU Parallel 20191122 ('Quantum Supremacy') released [stable]". parallel (Mailing list).
- "GNU Parallel". GNU.org.