Unread post Thu Oct 01, 2015 10:09 am

execution model question

Here's my understanding of how the execution model is set up:
At any given time, a chare "lives" withing the confines of a PE.
That PE runs a scheduler.
As such, the scheduler is in a position to enforce the following policy: no more than one method of a local chare can be executed by the PE at one given time. That being the case, if many callers invoke methods on a local chare, the PE schedules them for execution and they eventually get executed in an arbitrary sequence.