[Mpi3-rma] MPI-3.1 consideration slides

[Rajeev Thakur]

Looking at the exact sentence in the standard, 455:32-35, maybe it is already correct as it is. 

"A put or accumulate must not access a target window once a load/store update or a put or accumulate update to another (overlapping) target window has started on a location in the target window, until the update becomes visible in the public copy of the window.  ....., the restriction applies to operations even if they access disjoint locations in the window."

The overlapping in "A put or accumulate update to another (overlapping) target window" is probably intended there because if it is the same window, concurrent puts/accumulates to distinct parts of the same window are allowed (in the absence of local loads/stores).

Rajeev


On Dec 3, 2012, at 2:22 PM, Rajeev Thakur wrote:

> The change suggested on slide 2 is not right.
> 
> In the existing standard, for the separate memory model, the restriction is that a put/accumulate must not access a target window once a local update (to anywhere in the window) has started or a put/get (to *overlapping memory locations* in the window) has started.
> 
> You are allowed to do concurrent put/gets to distinct memory locations in the same window.
> 
> So the change needed is say "put/get to overlapping memory locations" instead of "put/get to an overlapping window".
> 
> Rajeev
> 
> 
> 
> On Dec 1, 2012, at 8:21 PM, Pavan Balaji wrote:
> 
>> All,
>> 
>> Please see the attached slides as consideration items for MPI-3.1.
>> 
>> -- Pavan
>> 
>> -- 
>> Pavan Balaji
>> http://www.mcs.anl.gov/~balaji
>> <mpi31-rma.pptx>_______________________________________________
>> mpi3-rma mailing list
>> mpi3-rma at lists.mpi-forum.org
>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
> 
> 
> _______________________________________________
> mpi3-rma mailing list
> mpi3-rma at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma