[Mpi3-rma] Updated MPI-3 RMA proposal 1

Underwood, Keith D keith.d.underwood at intel.com
Sun Jun 20 18:24:40 CDT 2010

> On 06/20/2010 06:09 PM, Rajeev Thakur wrote:
> >> The problem is that the definition of the "minimal set of
> >> things needed"
> >> is different for different people. That's why this definition
> >> is vague.
> >
> > One definition of inclusion in proposal 1: if feature X is not added,
> we
> > could as well just leave MPI-2 RMA as it is and go home.
> That's even more vague than what we had before :-).

How about "features in proposal 1 should be needed by all programming models".  And/or "features in proposal 1 should be necessary for implementing efficient one-sided applications".  Note that if it is relatively straightforward to build hardware/software/combo of the two to make a feature efficient, it wouldn't go in proposal 1 under the second definition.


More information about the mpiwg-rma mailing list