[Mpi3-hybridpm] First cut at slides
sko at cray.com
Thu Jun 20 17:20:42 CDT 2013
My apologies for some potentially dumb comments here, but I am missing
1. Can slides be expanded or can some one explain why the items in
question are desirable? (i.e. use cases) One case is libraries,
but I suspect there are others.
2. What does it mean if the first thread calls MPI_Init_thread with
required=MPI_THREAD_FUNNELD, then a later thread calls MPI_Init_thread
(MPI already initialized, now in "epoch" in slide-speak) with
required=MPI_THREAD_MULTIPLE? Does the second call see an error?
What does second call return for provided?
I suspect the second call succeeds and provided=MPI_THREAD_FUNNED
but I won't be surprised if that's wrong.
3. Slide 13 seems to imply that the implementation doesn't need to
actually implement an intermediate finalize/initialization, just
allow the calls. Is that correct? Does that satisfy the use
On 06/18/2013 07:47 AM, Jeff Squyres (jsquyres) wrote:
> For comment...
> Here's a first cut at slides that we discussed yesterday. It's meant to be a rollup of all the issues surrounding:
> - thread safe INIT / FINALIZE
> - reference-counting INIT / FINALIZE
> - nesting of INIT / FINALIZE
> I discovered a new issue when making up these slides: with ref-counting INIT/FINALIZE, since INIT/FINALIZE are collective, we have the same ref counts in all procs in MPI_COMM_WORLD. But the ref counts may be different in non-COMM_WORLD connected processes. What happens when they try to finalize over all connected processes? See slide 12 for an example. I'm not sure what the right answer is yet.
> Mpi3-hybridpm mailing list
> Mpi3-hybridpm at lists.mpi-forum.org
More information about the mpiwg-hybridpm