From goodell at [hidden] Thu Aug 21 15:02:55 2008 From: goodell at [hidden] (Dave Goodell) Date: Thu, 21 Aug 2008 15:02:55 -0500 Subject: [Mpi-22] new datatypes proposal on the wiki Message-ID: <381BE92B-2DB7-46D3-ACC7-92133D48D297@mcs.anl.gov> I have submitted a new 2.2 proposal to add several new datatypes to the C and C++ bindings, most of them in order to keep up with the C99 standard. This proposal also adds MPI_AINT and MPI_OFFSET as predefined datatypes. Comments are welcome. https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/mpi22/ NewPredefinedDatatypes -Dave Goodell Argonne National Laboratory From kannan.narasimhan at [hidden] Fri Aug 29 11:14:56 2008 From: kannan.narasimhan at [hidden] (Narasimhan, Kannan) Date: Fri, 29 Aug 2008 16:14:56 +0000 Subject: [Mpi-22] New proposal: Support for large message counts Message-ID: I have submitted a new 2.2 proposal to addresses the need for large message counts (i.e. counts greater than the size of 32-bit integer) for MPI calls that communicate messages. Please refer to https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/mpi22/LargeMsgCounts for details. Comments and suggestions are welcome. Thanx! Kannan From jsquyres at [hidden] Sun Aug 31 07:18:08 2008 From: jsquyres at [hidden] (Jeff Squyres) Date: Sun, 31 Aug 2008 13:18:08 +0100 Subject: [Mpi-22] new datatypes proposal on the wiki In-Reply-To: <381BE92B-2DB7-46D3-ACC7-92133D48D297@mcs.anl.gov> Message-ID: On Aug 21, 2008, at 9:02 PM, Dave Goodell wrote: > I have submitted a new 2.2 proposal to add several new datatypes to > the C and C++ bindings, most of them in order to keep up with the > C99 standard. This proposal also adds MPI_AINT and MPI_OFFSET as > predefined datatypes. Comments are welcome. > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/mpi22/NewPredefinedDatatypes Dave -- I like the idea of [u]int[x]_t types; that would certainly be helpful to some apps. I'm a little confused on your discussion about how bool and complex may be represented differently in each language. Are you saying that MPI_BOOL may be a different thing that [the already-existing] MPI::BOOL? Also, MPI_COMPLEX and MPI_DOUBLE_COMPLEX both already exist for the corresponding Fortran types (I can't remember offhand -- MPI_LONG_DOUBLE_COMPLEX may exist already, too). How do you propose to reconcile this conflict? -- Jeff Squyres Cisco Systems