In multiple instances with a runtime knowledge pattern, the number of multiple instances to be created (say, M) is determined before the first instance of the multiple instances gets initiated. The process moves to the subsequent tasks/activities only when all the tokens of the multiple instances (M) get completed, that is, all the instances of the set of multiple instances (M) get synchronized.

In static partial join patterns, the number of multiple instances to be created (say, M) is determined before the first instance of the multiple instances gets initiated. This is the same as in multiple instances with a prior runtime knowledge pattern; however, the process moves to subsequent tasks/activities only when N instances have been completed, where N is less than M. This pattern allows the process to continue only when a given number of instances have been completed; it's not necessary to wait for all the instances in the MultiInstance set...