I still disagree with the idea that MPI_INIT/FINALIZE is called per-MPI-Process. I know that this ticket does not deal with multiple MPI processes, but unless this sort of wording is stated correctly, ticket #310 won't work.

I also see confusing phraseology that implies that MPI-processes do things like "exit" and "make calls". I think it is going to be critical going forward (for ticket #310) that this sort of confusion is cleared up, not made worse. MPI Processes do not "exit" or "make calls".  Only OS Processes (threads within, actually) execute code and do things like "exit" and "make calls". The MPI processes are the TARGETS of code executed by the OS Process.

The big brick wall that ticket #310 will run into if we promote the idea that "every MPI process has its own MPI_INIT/FINALIZE" is that if there are multiple MPI processes then there is no clear way to initialize them since there is no argument to MPI_INIT/FINALIZE that specifies the MPI Process and there is no way to know what MPI process a thread might represent.


_______________________________________________
Douglas Miller                  BlueGene Messaging Development
IBM Corp., Rochester, MN USA                     Bldg 030-2 A401
dougmill@us.ibm.com               Douglas Miller/Rochester/IBM


Inactive hide details for Marc Snir ---02/01/2012 08:35:11 PM---Marc Snir <snir@mcs.anl.gov>Marc Snir ---02/01/2012 08:35:11 PM---Marc Snir <snir@mcs.anl.gov>

    Marc Snir <snir@mcs.anl.gov> 
    Sent by: mpi3-hybridpm-bounces@lists.mpi-forum.org

    02/01/2012 08:33 PM
    Please respond to
    mpi3-hybridpm@lists.mpi-forum.org


To

    mpi3-hybridpm@lists.mpi-forum.org,

cc


Subject

    [Mpi3-hybridpm] new text for MPI_INIT/MPI_FINALIZE

(See attached file: init-finalize.pdf)
_______________________________________________
Mpi3-hybridpm mailing list
Mpi3-hybridpm@lists.mpi-forum.org
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm