MPI_Cancel cannot simply be ignored. For example, if there is an MPI_Isend followed by MPI_Cancel, and the matching receive is never called, the cancel cannot fail. Rajeev

-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Greg Bronevetsky Sent: Monday, December 01, 2008 11:33 AM To: MPI-3 Collective Subgroup Discussions; MPI-3 Collective Subgroup Discussions Subject: Re: [mpi3-coll] Nonblocking collectives standard draft

...

However, if you all already think that MPI_Request_free is bad, then what do you think about the evil MPI_Cancel?

MPI_Cancel is a performance hint as far as I'm concerned, one that is frequently ignored. Can you point to any semantic difficulties with it?

Greg Bronevetsky Post-Doctoral Researcher 1028 Building 451 Lawrence Livermore National Lab (925) 424-5756 [email protected]

_______________________________________________ mpi3-coll mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-coll