Re: [OMPI users] Complexity of MPI_Comm_split and MPI_Comm_Create?

2015-01-20 Thread Jonathan Eckstein
Thanks very much, Edgar. I will definitely go with mpi_comm_create, then. I will need two extra scans to get set up for those operations, but that sounds much better than an Allgather (after some reflection, I was guessing that you would be doing an Allgather to organize the color keys). I

Re: [OMPI users] Complexity of MPI_Comm_split and MPI_Comm_Create?

2015-01-20 Thread Edgar Gabriel
Here are the communication operations occurring in the best case scenario in Open MPI right now: Comm_create: - Communicator ID allocation: 2 Allreduce operations per round of negotiations - 1 Allreduce operation for 'activating' the communicator Comm_split: - 1 Allgather operation fo