GNU Parallel caches stdout/stderr output from your jobs in temporary files in $TMPDIR.
If your jobs generate loads of output on stdout/stderr the files in $TMPDIR can grow big. With --compress you can ask GNU Parallel to compress them on the fly thus giving less disk I/O; and with --compress-program you can decide with program GNU Parallel should used for the compression. Give it a spin and provide some feedback. git clone git://git.savannah.gnu.org/parallel.git /Ole
