Re: [Mpi3-rma] nested locks
Agreed -- I think it's a bug and we should fix it asap. Best, Torsten On Mon, Aug 19, 2013 at 03:47:39PM -0500, Rajeev Thakur wrote:
It is still disallowed in MPI-3 by this sentence
pg 437, ln 28: Distinct access epochs for win at the same process must be disjoint.
On Aug 19, 2013, at 3:13 PM, Jim Dinan wrote:
Jeff,
With apologies if I have misunderstood your question. To clarify, the example Rajeev gave was forbidden in MPI-2. One could not have concurrent passive target access epochs, even if they accessed different targets. It is allowed in MPI-3.
~Jim.
On Mon, Aug 19, 2013 at 3:52 PM, Jeff Hammond <[email protected]> wrote: you mean lock at those ranks, not by those ranks? those aren't conflicting so i don't see the issue. maybe i am a dolt though.
jeff
On Mon, Aug 19, 2013 at 2:44 PM, Rajeev Thakur <[email protected]> wrote: Did we make any change in MPI 3.0 to allow nesting of MPI_Win_lock calls? For example,
Win_lock(rank 1) Win_lock(rank 2)
Win_unlock(rank 2) Win_unlock(rank 1)
I can't find the text we added or deleted to allow this.
But I do see text that disallows this: pg 437, ln 28: "Distinct access epochs for win at the same process must be disjoint."
Rajeev
_______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
-- Jeff Hammond [email protected]
_______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
_______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
_______________________________________________ mpi3-rma mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma
-- ### qreharg rug ebs fv crryF ----------- http://htor.inf.ethz.ch/ ----- Torsten Hoefler | Assistant Professor Dept. of Computer Science | ETH Z�rich Universit�tsstrasse 6 | Zurich-8092, Switzerland CAB E 64.1 | Phone: +41 76 309 79 29
participants (1)
-
Torsten Hoefler