Re: [OMPI users] OpenMPI and CUDA on cluster

2012-05-01 Thread Rohan Deshpande
I have solved this issue. All the paths were correct but I still had to use mpirun -x LD_LIBRARY_PATH while executing the job. Now works like a charm. Thanks for your responses. On Wed, May 2, 2012 at 4:04 AM, Trent wrote: > That is discussed on here: > > ** ** > > http://forums.nvidi

Re: [OMPI users] OpenMPI and CUDA on cluster

2012-05-01 Thread Trent
That is discussed on here: http://forums.nvidia.com/index.php?showtopic=227854 Maybe that could be your issue too. From: users-boun...@open-mpi.org [mailto:users-boun...@open-mpi.org] On Behalf Of Rohan Deshpande Sent: Tuesday, May 01, 2012 4:04 AM To: Open MPI Users Subject: [OMPI use

Re: [OMPI users] OpenMPI and CUDA on cluster

2012-05-01 Thread Jingcha Joba
Best way to find if the application binary can "see" the cuda library is to do ldd. On Tue, May 1, 2012 at 2:03 AM, Rohan Deshpande wrote: > Hi, > > I am trying to execute OpenMPI and CUDA code on a cluster. The code works > fine on single machine but when I try to execute it on cluster I get

Re: [OMPI users] Regarding mpi programming

2012-05-01 Thread Jeff Squyres
Unfortunately, that's not much of a description to go on to know where the problem is. In general, you should look to ensure that: a) you are sending exactly what you think you are sending (e.g., print out a sample value or two before and/or after you send) b) you are receiving exactly what y

Re: [OMPI users] OpenMPI and CUDA on cluster

2012-05-01 Thread Mohan L
On Tue, May 1, 2012 at 2:33 PM, Rohan Deshpande wrote: > Hi, > > I am trying to execute OpenMPI and CUDA code on a cluster. The code works > fine on single machine but when I try to execute it on cluster I get error: > > error while loading shared libraries: libcudart.so.4: cannot open shared > o

[OMPI users] OpenMPI and CUDA on cluster

2012-05-01 Thread Rohan Deshpande
Hi, I am trying to execute OpenMPI and CUDA code on a cluster. The code works fine on single machine but when I try to execute it on cluster I get error: error while loading shared libraries: libcudart.so.4: cannot open shared object file: No such file or directory I checked my PATH and LD_PATH

Re: [OMPI users] Serializing objects

2012-05-01 Thread Andreas Schäfer
Hi Omar, with serialization of C++ objects is fundamentally no different than sending structs with the C bindings. See [1] for an example of how to use MPI_Isend/MPI_Irecv. Be advised though that MPI won't guarantee that the communication will always run concurrently. If your objects contain poi