[OMPI users] adding library (libmpi_cxx.so.1)
Hi All, I have a problem to add a MPI library libmpi_cxx.so.1 to code which I am workign on it, I export the library to LD_LIBRARY_PATH. But whenever I do this, I get this error: Fatal error in MPI_Comm_dup: Invalid communicator, error stack: MPI_Comm_dup(168): MPI_Comm_dup(comm=0x0, new_comm=0x7fff39826eac) failed MPI_Comm_dup(96).: Invalid communicator It might be possible that I have several openmpi in my linux machine and the code could not recognize right one. I just have this library in these two path: /usr/lib64/openmpi/lib/ and /usr/local/lib/, but none of them is not working in my case. Installed openmpi in my machine is openmpi-1.5.4-3.fc16.x86_64 and I am working with Fedora. Does anybody can help me to figure out this problem? Thanks, Ami
Re: [OMPI users] adding library (libmpi_cxx.so.1)
This isn't quit enough information to figure out what's going on. Can you provide all the information listed here: http://www.open-mpi.org/community/help/ On Sep 13, 2013, at 7:43 PM, amirreza Hashemi wrote: > Hi All, > > I have a problem to add a MPI library libmpi_cxx.so.1 to code which I am > workign on it, I export the library to LD_LIBRARY_PATH. But whenever I do > this, I get this error: > Fatal error in MPI_Comm_dup: Invalid communicator, error stack: > MPI_Comm_dup(168): MPI_Comm_dup(comm=0x0, new_comm=0x7fff39826eac) failed > MPI_Comm_dup(96).: Invalid communicator > > It might be possible that I have several openmpi in my linux machine and the > code could not recognize right one. I just have this library in these two > path: /usr/lib64/openmpi/lib/ and /usr/local/lib/, but none of them is not > working in my case. Installed openmpi in my machine is > openmpi-1.5.4-3.fc16.x86_64 and I am working with Fedora. > Does anybody can help me to figure out this problem? > > Thanks, > Ami > ___ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
Re: [OMPI users] adding library (libmpi_cxx.so.1)
Jeff, Sorry about that, actually I am not sure what more exactly I can provide here, I am using openmpi 1.5.4-3.fc16.x86_64 with fedora machine, I installed that on yum install openmpi command. But when I am trying to make connection with some library libmpi_cxx.so.1 to main code, it gives me an error which it says there is no communicator Fatal error in MPI_Comm_dup: Invalid communicator, error stack: MPI_Comm_dup(168): MPI_Comm_dup(comm=0x0, new_comm=0x7fffde7fb24c) failed MPI_Comm_dup(96).: Invalid communicator Mpi libraries are found in two paths which are /usr/lib64/openmpi/lib/ and /usr/local/lib/ I would like to ask how can I can make connection to mpi library in fedora linux machine. It is indeed that I added libmpi_cxx.so.1 (libmpi_cxx.so.1 => /usr/lib64/openmpi/lib/libmpi_cxx.so.1 (0x7f120836e000)). I don't know is that enough information or not. I would be really apprecieated if you can help me to figure out this problem!! Thanks, Ami From: amirreza Hashemi To: "us...@open-mpi.org" Sent: Friday, September 13, 2013 1:43 PM Subject: adding library (libmpi_cxx.so.1) Hi All, I have a problem to add a MPI library libmpi_cxx.so.1 to code which I am workign on it, I export the library to LD_LIBRARY_PATH. But whenever I do this, I get this error: Fatal error in MPI_Comm_dup: Invalid communicator, error stack: MPI_Comm_dup(168): MPI_Comm_dup(comm=0x0, new_comm=0x7fff39826eac) failed MPI_Comm_dup(96).: Invalid communicator It might be possible that I have several openmpi in my linux machine and the code could not recognize right one. I just have this library in these two path: /usr/lib64/openmpi/lib/ and /usr/local/lib/, but none of them is not working in my case. Installed openmpi in my machine is openmpi-1.5.4-3.fc16.x86_64 and I am working with Fedora. Does anybody can help me to figure out this problem? Thanks, Ami
[OMPI users] any deadlocks in this sets of MPI_send and MPI_recv ?
Dear All, I have a question about using MPI_send and MPI_recv. *The object is as follows:* I would like to send an array Q from rank=1, N-1 to rank=0, and then rank 0 receives the Q from all other processors. Q will be put into a new array Y in rank 0. (of couse this is not realized by MPI). and then MPI_bcast is used (from rank 0) to broadcast the Y to all the processors. *Fortran Code is like:* if(myrank .eq. 0) then itag = myrank call mpi_send(Q.., 0, itag, .) else do i=1, N-1 itag = i call mpi_recv(QRECS..., i, itag, .) enddo endif call mpi_bcast(YVAR., 0, ..) *Problem I met is:* In my simulation, time marching is performed, These mpi_send and recv are fine for the first three time steps. However, for the fourth time step, the looping is only finished from i=1 to i=13, (totally 48 processors). That mean after 14th processors, the mpi_recv did not receive the date from them. Thus the code hangs there forever. Does deadlock occur for this situation? How can I figure out this problem? Thank you so much if anyone can give me some suggestions. best regards, Huangwei
Re: [OMPI users] any deadlocks in this sets of MPI_send and MPI_recv ?
*The code I would like to post is like this:* * * if(myrank .ne. 0) then itag = myrank call mpi_send(Q.., 0, itag, .) else do i=1, N-1 itag = i call mpi_recv(QRECS..., i, itag, .) enddo endif call mpi_bcast(YVAR., 0, ..) best regards, Huangwei On 13 September 2013 23:25, Huangwei wrote: > Dear All, > > I have a question about using MPI_send and MPI_recv. > > *The object is as follows:* > I would like to send an array Q from rank=1, N-1 to rank=0, and then rank > 0 receives the Q from all other processors. Q will be put into a new array > Y in rank 0. (of couse this is not realized by MPI). and then MPI_bcast is > used (from rank 0) to broadcast the Y to all the processors. > > *Fortran Code is like:* > if(myrank .eq. 0) then > itag = myrank > call mpi_send(Q.., 0, itag, .) > else > do i=1, N-1 > itag = i > call mpi_recv(QRECS..., i, itag, .) > enddo > > endif > > call mpi_bcast(YVAR., 0, ..) > > *Problem I met is:* > In my simulation, time marching is performed, These mpi_send and recv are > fine for the first three time steps. However, for the fourth time step, the > looping is only finished from i=1 to i=13, (totally 48 processors). That > mean after 14th processors, the mpi_recv did not receive the date from > them. Thus the code hangs there forever. Does deadlock occur for this > situation? How can I figure out this problem? > > Thank you so much if anyone can give me some suggestions. > > best regards, > Huangwei > > > > >
