You should make individual jobs for all machines. This way you know immediately which jobs failed, and not just a main job with multiple hosts in it. It's a good practice to do this, a recommended practice by some experts in the field, and something I've been getting used to for a while now, and liking thus far. Of course you need a good naming convention too, so you know what is what.
Later you can run just the individual host job, and not the whole bunch, if you were to gather statistics on pass vs failure, doing it my way, you'll have a better success rate in terms of percentage. Auditors love this, as do managers. ;)