[mpich-discuss] How expensive is MPI_Win_create() compared to memcpy()?

[Dave Goodell]

On Dec 28, 2011, at 6:19 AM CST, Jed Brown wrote:

> On Tue, Dec 27, 2011 at 22:55, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>> I see.  In that case, the dynamically attaching memory to windows option in MPI-3 is what you need.  In this case, the window object is created collectively once, and then memory can be attached to it as a local operation without requiring collective synchronization.
> 
> Perfect, let me know when you are ready for someone to test an implementation. I can make any of these variants into run-time options that become available if the MPI provides the necessary functions, so it will be an easy test environment.

Jed and I previously talked about this offline.  I don't think that dynamic windows are a good match for his problem because all that it would do is shift the burden of determining buffer address starting offsets from the MPI implementation to his code.

-Dave