This is exactly the problem. If you run with <10 processes, it looks like you're sending more than 100 int's to each process, and the message gets truncated at the receiver. MPI defines that this...