[mpiwg-hybridpm] Reminder: Hybrid WG telecon tomorrow

Jeff Hammond jeff.science at gmail.com
Mon Feb 10 11:51:40 CST 2014

> For discussion today: How to deal with MPI_Comm_compare(ep_comm[0],
> ep_comm[1], &result).  Where ep_comm is an array of handles to different
> ranks in the same endpoints comm.  Currently, MPI defines: IDENT, CONGRUENT,
> SIMILAR, and UNEQUAL.  None are a good match, seems like we need a new one
> for the case where they are the same communicator and same group (IDENT),
> but represent a different rank in the comm.

I suggest MPI_ALIASED to describe this situation.


Jeff Hammond
jeff.science at gmail.com

More information about the mpiwg-hybridpm mailing list