<html>
<body>
ditto here (we are all busy)<br><br>
it may be a short meeting then!<br><br>
At 09:55 AM 12/12/2008, you wrote:<br>
<blockquote type=cite class=cite cite="">I am sorry Jeff, there was no
progress on my front wrt ABI.<br>
<br>
<b>From:</b> mpi3-abi-bounces@lists.mpi-forum.org
[<a href="mailto:mpi3-abi-bounces@lists.mpi-forum.org" eudora="autourl">
mailto:mpi3-abi-bounces@lists.mpi-forum.org</a>] <b>On Behalf Of </b>Jeff
Brown<br>
<b>Sent:</b> Thursday, December 11, 2008 9:21 AM<br>
<b>To:</b> MPI 3.0 ABI working group<br>
<b>Cc:</b> craig Rasmussen; ddd@lanl.gov; Ralph Castain<br>
<b>Subject:</b> Re: [Mpi3-abi] Open MPI and ABI<br>
<br>
All,<br><br>
At the last meeting I attended (Dublin) we discussed developing a
prototype "morph" ABI layer with initial implementation
targeting the December meeting. Let's take stock of this effort
when we meet next week. As I recall, Ezra offered to lead the
windows implementation, and Alexander linux. This exercise is meant
to expose some of the implementation details and help drive us toward a
detailed standard proposal we can take to the larger group.<br><br>
Ezra and Alexander, please brief us on this effort when we meet next
Wednesday.<br><br>
>From the Dublin meeting:<br>
•MPI ABI standard, separate from MPI API standard, but under the umbrella
of the MPI Forum<br>
•Dynamic link is the main objective (binary compatibility)<br>
•Startup (mpirun/mpiexec) is a secondary objective<br>
•Include all major language bindings in the ABI standard (C, C++,
Fortran) as we go<br>
•Tied to MPI 2.1 and forward<br>
•Implementers may choose to comply (or not)<br>
•May be implemented as a morph layer or native<br>
•Initial reference implementation will be a morph layer (tbd who
provides)<br>
•May be different per platform (Linux, Windows, …)<br>
•Different voting rules separate from MPI API forum (?)<br>
•Need participation from major implementors in order to make
progress<br><br>
mpi.h: MPICH 2 (1.0.7) as base with mods as needed (e.g. max values for
compile time values used in declarations)<br>
Proceed with a reference morph layer implementation targeting (at least)
MPICH 2.0, OpenMPI <br>
Load a different module and it just works<br>
Demonstrate with a ring program<br>
Defer startup consideration<br><br>
Proof of concept reference morph layer implementations:<br>
•Flesh out details by writing some code<br>
•Linux on X86-64 cluster Red Hat Linux/SUSE<br>
•Windows on X86-64 cluster<br>
•Test against ring program<br>
•<b>Bring back to the December forum meeting<br><br>
</b>see ya next week<br><br>
Jeff<br>
_______________________________________________<br>
mpi3-abi mailing list<br>
mpi3-abi@lists.mpi-forum.org<br>
<a href="http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi" eudora="autourl">
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi</a></blockquote>
</body>
</html>