In developing large scale simulations in workgen, we sometimes want to have a set of worker (test) threads to synchronize some activity.  For example, consider 1000 threads that each sleep a second and then do 1000 operations, repeatedly. They can start off with a synchronized heavy load (a million operations), followed by a lull, and then continuing.  Except that as time marches on, the threads will naturally complete their thousand operations at different times, and thus get out of sync. The initial test conditions evolve into something quite different.  If the desire is to continue with a set of threads in a synchronized manner, there is no current way to do that.