Hello,
I was running the pi demo on multicores (Pentium Dual, Core i7) to see
it scale, but sometimes the time measurements return disparate results.
The FAQ suggests processor affinity as one possible reason for that. For
instance, the demo takes 3s in one core of the Pentium Dual
$ mpirun -h
Are you running on nodes with both MX and OpenFabrics?
I don't know if this is a well-tested scenario -- there may be some strange
interactions in the registered memory management between MX and OpenFabrics
verbs.
FWIW, you should be able to disable Open MPI's memory management at run time in
Hi all,
I have a question regarding the support for heterogeneous systems. Using the
latest version of OpenMPI, is it possible to have a single MPI application
running across a cluster of mixed operating systems, say for example Windows
and Linux?
Thanks,
Hicham
Shiqing --
Do we support mixed Windows + Linux jobs?
On Jun 1, 2010, at 1:17 PM, Hicham Lahlou wrote:
> Hi all,
>
> I have a question regarding the support for heterogeneous systems. Using the
> latest version of OpenMPI, is it possible to have a single MPI application
> running across a clu
Hi,
Unfortunately, we don't have such support at moment.
Regards,
Shiqing
On 2010-6-1 8:49 PM, Jeff Squyres wrote:
Shiqing --
Do we support mixed Windows + Linux jobs?
On Jun 1, 2010, at 1:17 PM, Hicham Lahlou wrote:
Hi all,
I have a question regarding the support for heterogeneou
After patching, I get:
make[3]: Entering directory
`/scr_iter/cary/facetspkgs/builds/openmpi-1.4.2/nodl/ompi/contrib/vt/vt'
make[3]: *** No rule to make target
`/scr_iter/cary/facetspkgs/builds/openmpi/ompi/contrib/vt/vt/m4/acinclude.compinst.m4',
needed by
`/scr_iter/cary/facetspkgs/builds/o
