I think that Pavan's question was motivated by the fact that ARMCI seems to allow this. However, just because ARMCI does this does not make it right. If a user reported this as a problem with an MPI-compliant ARMCI implementation, I would tell him or her to write a better application. NWChem definitely does not need this. I suspect that the only reason ARMCI allows this is because they didn't have a reason to prohibit it, but it requires a s�ance to know for sure :-( Jeff On Thu, Apr 28, 2011 at 1:08 PM, William Gropp <[email protected]> wrote:
I agree - I don't think this is an issue for users, including middleware, of the RMA, and I would not want it to be different than the send/recv case.
Bill
On Apr 28, 2011, at 7:57 AM, Torsten Hoefler wrote:
On Thu, Apr 28, 2011 at 12:36:39AM -0500, Pavan Balaji wrote:
Do these need to be relaxed in MPI-3 as well to make them undefined for Put/Get? It might be OK to make it undefined in Accumulate as well.
I don't think so because it's also erroneous for send/recv.
All the Best, �Torsten
-- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Torsten Hoefler � � � � | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street � � | Urbana, IL, 61801 NCSA Building � � � � � | +01 (217) 244-7736 _______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign
_______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
-- Jeff Hammond Argonne Leadership Computing Facility [email protected] / (630) 252-5381 http://www.linkedin.com/in/jeffhammond