[Mpi3-hybridpm] Endpoints Proposal

Pavan Balaji balaji at mcs.anl.gov
Wed Mar 20 15:21:41 CDT 2013


Jeff,

On 03/20/2013 03:10 PM US Central Time, Jeff Hammond wrote:
> Yes, I am thinking about endpoints generally, not the specific
> proposal on the table, which I see as a being one good but not
> necessarily exclusive solution for the MPI-UPC issues, as others have
> clearly noted.

We are not excluding other use cases.  The reason the interoperability
one is important is because we cannot do it with MPI currently.
Performance optimizations are important too, of course, but whether or
not we can get the same improvements in current is unclear.

> I hope that a second proposal will be able to address concurrency
> issues that I have with MPI, but I recognize that this is more
> controversial.  I think I have a more general solution to this problem
> but it will take me some time to document properly.

Sounds great.  Other proposals that address this problem are most welcome.

 -- Pavan

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji



More information about the mpiwg-hybridpm mailing list