Fair enough. If we want to have info arguments to communicator constructors moving forward (as MPI_Comm_dup_info implies), then we should add the argument to this routine. ~Jim. On 7/18/12 2:45 PM, William Gropp wrote:
I suspect the info hint
nowildcards=true
would be useful for some implementations. Note also the MPI_Comm_dup_info call; its not just split_type.
Bill
William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign
On Jul 18, 2012, at 2:43 PM, MPI Forum wrote:
#286: Noncollective Communicator Creation -------------------------------------+------------------------------------- Reporter: jdinan | Owner: Type: New | jdinan routine(s) | Status: Priority: Passed | assigned Version: MPI | Milestone: 3.0 | 2012/05/28 Japan Keywords: | Resolution: communicators | Implementation status: Author: Bill Gropp: 0 | Completed Author: Adam Moody: 0 | Author: Rich Graham: 0 Author: Dick Treumann: 0 | Author: Torsten Hoefler: 0 Author: George Bosilca: 0 | Author: Jesper Larsson Traeff: 0 Author: Bronis de Supinski: 0 | Author: David Solt: 0 Author: Jeff Squyres: 0 | Author: Rajeev Thakur: 0 Author: Rolf Rabenseifner: 0 | Author: Alexander Supalov: 0 -------------------------------------+-------------------------------------
Comment (by jdinan):
I don't think we ever had a use case for the info argument, so we never added it. The info key in MPI_Comm_split_type was added to allow the user to specify machine topology hints.
-- Ticket URL: <https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/286#comment:16> MPI Forum <https://svn.mpi-forum.org/> MPI Forum
_______________________________________________ mpi-forum mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum