below. This doesn’t validate the one-sided
run thou.
George.
> On Apr 14, 2015, at 11:07 , MOHR STEPHAN 239883 wrote:
>
> Hi Nathan,
>
> I tried with send/recv, but the outcome is the same. It works for small
> counts (e.g. n=100), but hangs for larger ones (e.g. n=1000
you
also give master a try?
-Nathan
On Tue, Apr 14, 2015 at 06:43:31AM +, MOHR STEPHAN 239883 wrote:
>Hi Howard,
>
>I tried with 1.8.5rc1, but it doesn't work either.
>
>The output of ompi_info is attached.
>
&g
www.open-mpi.org/software/ompi/v1.8/downloads/openmpi-1.8.5rc1.tar.gz>?
The test appears to work without problem using mpich, at least with 4 ranks.
Thanks,
Howard
2015-04-13 10:40 GMT-06:00 MOHR STEPHAN 239883
mailto:stephan.m...@cea.fr>>:
Hi there,
I've got an issue when usi
Hi there,
I've got an issue when using a derived data type created by
mpi_type_create_struct in a one-sided communication.
The problem can be reproduced using the small standalone program which I
attached. It just creates a type which should be equivalent to n contiguous
elements. This type is
