[mpiwg-rma] schedule telecon

Jeff Hammond jeff.science at gmail.com
Sat Nov 2 17:14:03 CDT 2013


So basically, we need a 3-4 hour telecon? :-)

Jeff

Sent from my iPhone

On Nov 2, 2013, at 4:35 PM, Pavan Balaji <balaji at mcs.anl.gov> wrote:

>
> Some additional things to discuss:
>
> 1. non-interleaved Ops and attach/detach info argument —- if the application might interleave attach/detach with PUT/GET operations, the MPI implementation cannot cache things like the memory registration information for the target buffer.  This info hint can help in such cases.
>
> 2. WIN_SYNC semantics in unified windows.  We had some discussion on this over email, but this needs to be finalized.  This is needed for ordering.  I think the wording we agreed on was to say that the public/private windows are disjoint, but will eventually be consistent in UNIFIED.  So for overlapping memory regions, UNIFIED and SEPARATE are not too different for portable programs.  UNIFIED has richer capabilities for nonoverlapping memory regions.
>
> 3. I want to initiate some discussion on threads+RMA.  We might need some discussion on this, as this would mean that each thread has its own private window.  That is, load/store from one might not be visible in another without a WIN_SYNC.
>
> 4. I want to initiate some discussion on exposing memory barrier semantics in MPI.  We are kind of half-exposing it with WIN_SYNC in UNIFIED.  If we do this, I’d recommend that we expose WIN_SYNC_LOAD and WIN_SYNC_STORE as well for completeness.  I expect this to be highly debated, of course; I just want to get this item on the agenda.
>
>  —- Pavan
>
> On Nov 2, 2013, at 4:12 PM, Jeff Hammond <jeff.science at gmail.com> wrote:
>
>> I would like to schedule a telecon for the RMA WG to discuss the
>> following three tickets.  We've discussed them all over email but I
>> want to get more conclusive feedback before the Chicago meeting.
>>
>> - extend the use of MPI_WIN_SHARED_QUERY to all windows
>> (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/397)
>> - request-based remote completion for RMA
>> (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/398)
>> - generalize same_op_no_op and allow user to specify all ops to be
>> used (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/399)
>>
>> Thanks,
>>
>> Jeff
>>
>> --
>> Jeff Hammond
>> jeff.science at gmail.com
>> _______________________________________________
>> mpiwg-rma mailing list
>> mpiwg-rma at lists.mpi-forum.org
>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-rma
>
> --
> Pavan Balaji
> http://www.mcs.anl.gov/~balaji
>
> _______________________________________________
> mpiwg-rma mailing list
> mpiwg-rma at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-rma



More information about the mpiwg-rma mailing list