Comment 4 for bug 2061166

Faster machines spend less time executing hooks, so that might make sense.

I think we need to tweak the way we wait for a model to be done slightly.