Re: [OMPI users] Unable to compile for libnumactl and libnumactl-devel

2015-11-03 Thread Fabian Wein
On 11/03/2015 06:13 PM, Dave Love wrote: Fabian Wein writes: There is an old OpenFOAM installation which includes and old open-mpi, this might cause the trouble. OpenFOAM should definitely be built against the system MPI (and, in general, you should avoid bundled libraries wherever possible

Re: [OMPI users] Setting bind-to none as default via environment?

2015-11-03 Thread Nick Papior
This is what I do to successfully get the best performance for my application using OpenMP and OpenMPI: (note this is for 8 cores per socket) mpirun -x OMP_PROC_BIND=true --report-bindings -x OMP_NUM_THREADS=8 --map-by ppr:1:socket:pe=8 It assigns 8 cores per MPI processor separated by sockets,

Re: [OMPI users] Setting bind-to none as default via environment?

2015-11-03 Thread Dave Love
Lloyd Brown writes: > No problem. It wasn't much of a delay. > > The scenario involves a combination of MPI and OpenMP (or other > threading scheme). Basically, the software will launch one or more > processes via MPI, which then spawn threads to do the work. > > What we've been seeing is that,

Re: [OMPI users] Ubuntu/Debian packages for recent version (for Travis CI support)

2015-11-03 Thread Dave Love
Jeff Hammond writes: > I setup Travis CI support for ARMCI-MPI but the available version in > whatever Ubuntu they use is buggy. For example: > https://travis-ci.org/jeffhammond/armci-mpi/jobs/0279. [I can't make sense of that to see what the problem actually is.] > I have not checked late

Re: [OMPI users] Unable to compile for libnumactl and libnumactl-devel

2015-11-03 Thread Dave Love
Fabian Wein writes: > There is an old OpenFOAM installation which includes and old open-mpi, > this might > cause the trouble. OpenFOAM should definitely be built against the system MPI (and, in general, you should avoid bundled libraries wherever possible IMHO). > I also suspect that sourcing

Re: [OMPI users] Setting bind-to none as default via environment?

2015-11-03 Thread Lloyd Brown
No problem. It wasn't much of a delay. The scenario involves a combination of MPI and OpenMP (or other threading scheme). Basically, the software will launch one or more processes via MPI, which then spawn threads to do the work. What we've been seeing is that, without something like '--bind-to

Re: [OMPI users] Setting bind-to none as default via environment?

2015-11-03 Thread Ralph Castain
Sorry for delay - was on travel. hwloc_base_binding_policy=none Alternatively, you may get better performance if you bind to numa or socket levels, assuming you want one proc per socket: hwloc_base_binding_policy=socket [or numa] rmaps_base_mapping_policy=socket [or numa] HTH Ralph > On Nov 2