The responsible programmer for this code has conceded the point and we
will be replacing all offending examples with the MPI_IN_PLACE solution.
Thanks for the input.
T. Rosmond
On Thu, 2010-09-16 at 13:56 -0700, Tim Prince wrote:
> On 9/16/2010 9:58 AM, David Zhang wrote:
> > It's compiler spe
On 9/16/2010 9:58 AM, David Zhang wrote:
It's compiler specific I think. I've done this with OpenMPI no
problem, however on one another cluster with ifort I've gotten error
messages about not using MPI_IN_PLACE. So I think if it compiles, it
should work fine.
On Thu, Sep 16, 2010 at 10:01
Tony
You are depending on luck. The MPI Standard allows the implementation to
assume that send and recv buffers are distinct unless MPI_IN_PLACE is
used. Any MPI implementation may have more than one algorithm for a given
MPI collective communication operation and the policy for switching
al
The description for MPI_GATHERV says (from
http://www.mpi-forum.org/docs/mpi22-report/node95.htm#Node95):
The specification of counts, types, and displacements should not cause any
location on the root to be written more than once. Such a call is erroneous.
The ``in place'' option for intracomm
It's compiler specific I think. I've done this with OpenMPI no problem,
however on one another cluster with ifort I've gotten error messages about
not using MPI_IN_PLACE. So I think if it compiles, it should work fine.
On Thu, Sep 16, 2010 at 10:01 AM, Tom Rosmond wrote:
> I am working with a
I am working with a Fortran 90 code with many MPI calls like this:
call mpi_gatherv(x,nsize(rank+1),
mpi_real,x,nsize,nstep,mpi_real,root,mpi_comm_world,mstat)
'x' is allocated on root to be large enough to hold the results of the
gather, other arrays and parameters are defined correctly, an
