Hi,
The calls to broadcast in qmmm_module.F90 employ overloading via
the interface statements below; basically, the interface statements
define explicit instances of the name broadcast (the type signatures
are those of the subroutines specified in the module procedure
statements), and the actual instance in a call broadcast statement
is resolved by the argument type(s).
./qmmm_nml_module.F90:298: interface broadcast
./qmmm_vsolv_module.F90:177: interface broadcast
./qmmm_qmtheorymodule.F90:49: interface Broadcast
./qmmm_struct_module.F90:178: interface broadcast
In general, this all looks correct, albeit perhaps overly fancy.
scott
On Tue, Jan 10, 2012 at 08:01:51AM -0500, Jason Swails wrote:
> Linux x86_64: (Ubuntu 10.04) Intel 12.1.0 openMPI
> Linux x86_64: (Ubuntu 10.04) Intel 12.1.0 mpich2
>
> Mac OS X 10.7: GNU 4.4.6 mpich2
>
> It happens on all of the above. I can show you today.
>
> On Tue, Jan 10, 2012 at 7:46 AM, Taisung Lee <taisung.biomaps.rutgers.edu>wrote:
> > The current build runs fine on my machine--FC 13; OpenMPI 1.4.3
> > (x86_64); gnu compilers.
> >
> > Could you post your OS and OpenMPI versions?
> >
> > > -----Original Message-----
> > > From: Jason Swails [mailto:jason.swails.gmail.com]
> > >
> > > All of the QM/MM tests are segfaulting for me in parallel (Mac OS X and
> > Linux, both
> > > GNU and Intel, mpich2 and openMPI). Is anyone else seeing that?
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Jan 10 2012 - 08:30:02 PST
