See https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/357. It's not a proposal in the sense that I do not have a PDF demonstrating a LaTeX implementation of the suggested text changes, but I did provide in my ticket the text changes I think address this issue. Jeff On Fri, Dec 16, 2011 at 10:38 AM, Jeff Squyres <[email protected]> wrote:
+1 / good points.
Will you make a proposal about this?
On Dec 16, 2011, at 10:44 AM, Jeff Hammond wrote:
Are MPI_Initialized and MPI_Query_thread guaranteed to be thread-safe by all implementations? If these MPI calls cannot be made in a fully multithreaded context, how can a multithreaded library determine that it can make MPI calls in an unrestricted fashion, i.e. determine if MPI was initialized with MPI_THREAD_MULTIPLE? Must the multithreaded library mutex around these two MPI calls and assume that MPI_THREAD_SERIALIZED is always available?
Can MPI_Is_thread_main be called from a thread that is not main when only MPI_THREAD_FUNNELED is supported? I recall that the funneled mode stipulates that only the main thread can make MPI calls and there is no text excepting this routine from that constraint.
Thanks,
Jeff
-- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute [email protected] / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki-old.alcf.anl.gov/index.php/User:Jhammond _______________________________________________ mpi-forum mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
-- Jeff Squyres [email protected] For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
_______________________________________________ mpi-forum mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
-- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute [email protected] / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond