[eindiran]

The value add is that you don't need to do the `xargs --max-procs N`, yourself. By default N is 1 for xargs. For parallel, the default is N = number of CPUs.

Additionally, you can run a series of unrelated commands that aren't from a list/piped in with parallel using the `--` syntax:

`parallel -j 3 -- ls df "echo hi"`

You can limit system load using parallel, which as far as I know isn't possible with xargs: `parallel -l L` where L is the average system load you want to remain beneath.