Re: [OMPI users] Problems using Open MPI 1.8.4 OSHMEM on Intel Xeon Phi/MIC

Ralph, now i remember this part ... IIRC, LD_LIBRARY_PATH was never forwarded when remote starting orted. i simply avoided this issue by using gnu compilers, or gcc/g++/ifort if i need intel fortran /* you already mentionned this is not officially supported by Intel */ What about adding a new

Re: [OMPI users] Problems using Open MPI 1.8.4 OSHMEM on Intel Xeon Phi/MIC

Gilles and Ralph, thanks! $ shmemrun -H mic0,mic1 -n 2 -x SHMEM_SYMMETRIC_HEAP_SIZE=1M $PWD/mic.out [atl1-01-mic0:192474] [[29886,0],0] ORTE_ERROR_LOG: Not found in file base/plm_base_launch_support.c at line 440 Hello World from process 0 of 2 Hello World fr

Re: [OMPI users] MPI_Init() time

As Ralph mentioned, the 1.4.x series is very old. Any chance you can upgrade to 1.8.x? > On Apr 15, 2015, at 7:12 AM, cristian wrote: > > Hello, > > I noticed when performing a profiling of an application that the MPI_init() > function takes a considerable amount of time. > There is a big d

Re: [OMPI users] Why 3 threads per MPI task in pure MPI job?

Those are async progress threads - they block unless something requires doing > On Apr 15, 2015, at 8:36 AM, Sasso, John (GE Power & Water, Non-GE) > wrote: > > I stumbled upon something while using ‘ps –eFL’ to view threads of processes, > and Google searches have failed to answer my questio

[OMPI users] Why 3 threads per MPI task in pure MPI job?

I stumbled upon something while using 'ps -eFL' to view threads of processes, and Google searches have failed to answer my question. This question holds for OpenMPI 1.6.x and even OpenMPI 1.4.x. For a program which is pure MPI (built and run using OpenMPI) and does not implement Pthreads or Op

Re: [OMPI users] Select a card in a multi card system

Hi, With MXM, you can specify list of devices to use for communication: -x MXM_IB_PORTS="mlx5_1:1,mlx4_1:1" also select specific or all transpoirts: -x MXM_TLS=shm,self,ud To change port rate one can use *ibportstate* *http://www.hpcadvisorycouncil.com/events/2011/switzerland_workshop/pdf/Pres

Re: [OMPI users] MPI_Init() time

With an OMPI that old, it’s anyone’s guess - I have no idea. > On Apr 15, 2015, at 4:12 AM, cristian wrote: > > Hello, > > I noticed when performing a profiling of an application that the MPI_init() > function takes a considerable amount of time. > There is a big difference when running 32 p

Re: [OMPI users] mpi_type_create_struct not working for large counts

Stop using mpif.h. Upgrading to use the "mpi" module is a very simple upgrade, and gives you at least some measure of type safety and argument checking (e.g., it would have failed to compile with that missing ierror argument). If you have a modern enough Fortran compiler (e.g., gfortran >= 4.

[OMPI users] MPI_Init() time

Hello, I noticed when performing a profiling of an application that the MPI_init() function takes a considerable amount of time. There is a big difference when running 32 processes over 32 machines and 32 processes over 8 machines (Each machine has 8 cores). These are the results of the profil

[OMPI users] Select a card in a multi card system

If you have a system with two IB cards, can you choose using a command line switch which card to use with Openmpi? Also a more general question - can you change (or throttle back) the speed at which an Infiniband card works at? For example, to use an fDR card at QDR speeds. Thanks for any insight

Re: [OMPI users] mpi_type_create_struct not working for large counts

Hi George, I think you forgot the ierror argument in the call to mpi_irecv, but after correcting this it works fine. Thanks a lot for pointing out the issue of the eager limit! But as you said, this does not directly solve my one-sided problem... Thanks, Stephan ___