[petsc-users] Issue with window SF type using derived datatypes for reduction

[Lawrence Mitchell]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 10/02/15 01:31, Jed Brown wrote:
> Lawrence Mitchell <lawrence.mitchell at imperial.ac.uk> writes:
>> Having just tried a build with --download-mpich, I notice this
>> problem does not occur.  So should I shout at the OpenMPI team?
> 
> Open MPI has many long-standing bugs with one-sided and datatypes.
> I have pleaded with them to error instead of corrupting memory or 
> returning wrong results for unsupported cases.  My recommendation
> is to not use -sf_type window with Open MPI.
> 
> I hear that they have newfound interest in fixing the decade-old 
> one-sided bugs, so I would say it is worth reporting this issue.

Done, and seemingly fixed, in https://github.com/open-mpi/ompi/issues/385

I shall see if any more issues arise.

Cheers,

Lawrence


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1

iQEcBAEBAgAGBQJU41T4AAoJECOc1kQ8PEYv3/cH/0/Q1tLsf1et4EVXE4hPOgOs
1EfqELCuX+qtSO0obGeZ4n7x1ixGAB2l0OnHCv4qXOp57pT2WNAFtMlCizZSsBTX
ekGdwWJIIV99Qh8NdnaGBKTZLA1DWrgTueIzmdAJrPSmU5DU8kWrrGW0Qr7nwtwn
zBCc2iQXCwJLgEDKIhIQh9uPrNFWa4IQohD/9UFrD/TetT5CEdYrAvnxhcirz1qX
3R3jbFTB11slQlJ/txRqJNDhlRlSAI5mtRxwMRDG+lI/UF1I782kR8ClXOODjHJ+
eOVqQh4rQ3atRlJ2ynr7RB86BWHxM9Ktl+nGGUQyT1W+07DoUZ+UrctLT4gecws=
=KP++
-----END PGP SIGNATURE-----