Hi there,<br><br>i have a big problem by running MPICH2 programs which use the Boost.MPI library. When i'm trying to run programs on <b>more </b>than one node, collective operations like <code></code><a href="http://boost.org/doc/libs/1_44_0/doc/html/boost/mpi/communicator.html#id918378-bb" target="_blank">communicator::barrier</a>, <code></code><a href="http://boost.org/doc/libs/1_44_0/doc/html/boost/mpi/broadcast.html" title="Function broadcast" target="_blank">broadcast,</a> or even the <code><a href="http://boost.org/doc/libs/1_44_0/doc/html/boost/mpi/environment.html" title="Class environment" target="_blank">environment</a></code>
destructor (cause of FINALIZE, which is colletive) causing the programm to crash. Maybe its a problem of Boost and the communication cystem, i use ch3:nemesis.<br>My errors are like this :<br><br><b>[1]terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::mpi::exception> >'<br>
<br>[1] what(): MPI_Barrier: Other MPI error, error stack:<br>[1]PMPI_Barrier(362).................: MPI_Barrier(MPI_COMM_WORLD) failed<br>[1]MPIR_Barrier_impl(255)............: <br>[1]MPIR_Barrier_intra(79)............: <br>
[1]MPIC_Sendrecv(186)................: <br>[1]MPIC_Wait(534)....................: <br>[1]MPIDI_CH3I_Progress(184)..........: <br>[1]MPID_nem_mpich2_blocking_recv(895): <br>[1]MPID_nem_tcp_connpoll(1746).......: Communication error with rank 0: </b><br>
<br>I also tested this with the simple broadcast example from the Boost.MPI tutorial - same errors..<br>But when using the original MPI equivalent without the Boost.MPI library, such as <a href="http://www.mpi-forum.org/docs/mpi-11-html/node66.html#Node66" target="_blank"><code><span>MPI_Barrier</span></code></a>, the programm runs well. I am using MPICH2 on Ubuntu 10.04 x86 platforms.<br>
Someone had problems like this, or know a fix for that?<br><br>Regards,<br><font color="#888888"><br>Stephan</font>