[mpiwg-rma] same_op_no_op

Jeff Hammond jeff.science at gmail.com
Thu Mar 13 16:08:45 CDT 2014

On Thu, Mar 13, 2014 at 2:52 PM, Balaji, Pavan <balaji at anl.gov> wrote:
> On Mar 13, 2014, at 2:47 PM, Jeff Hammond <jeff.science at gmail.com> wrote:
>> If we thought it was a mistake in MPI-2 to disallow it, wouldn't we
>> have tried to relax it in MPI-3?  Brian had commented before that we
>> might have been confused when defining same_op_no_op and really meant
>> same_op_no_op_replace, which is a perfectly reasonable default
>> semantic.
> It was not disallowed in MPI-2.  I meant that we made a mistake in MPI-3 to disallow it since that is not backward compatible.

You told me two days ago it was at best undefined, which is no more
useful than disallowed:
On Tue, Mar 11, 2014 at 12:04 AM, Balaji, Pavan <balaji at anl.gov> wrote:
> MPI-2 defines atomicity only for the same operation, not any operation for MPI_ACCUMULATE.
>   — Pavan

> IMO, we should have kept the same semantics as MPI-2, but allowed the user to relax it with info arguments.

Well we broke backwards compatibility but made it almost impossible
for anyone to notice and certainly didn't add an advice to users so we
are clearly all jerks.


Jeff Hammond
jeff.science at gmail.com

More information about the mpiwg-rma mailing list