With lots of help I think I am getting close to a "hello, world" BGL, MPI example to execute. Not sure about the error I am getting from execution from libs/mpi/example on symmetric multiprocessing (SMP) cluster of compute nodes, 1. compilation/translation of hello_world.cpp, works :-) $ mpiCC -I ../boost_1_48_0 \ -I ../boost_1_48_0/boost/mpi \ -I ../boost_1_48_0/stage/lib \ -lboost_mpi \ -lboost_thread \ -lboost_regex \ -lboost_graph -lboost_graph_parallel \ -lboost_serialization \ -lpthread \ hello_world.cpp -o hello_world 2. execution, error :-( $ mpirun -np 2 ./hello_word *** Process received signal *** Signal: Segmentation fault (11) *** Signal code: Address not mapped (1) *** Failing at address (nil) *** /lib64/libpthread.so.0 *** mpirun noticed process rank 0 with PID 16684 Maybe someone has seen this before and knows of a fix. Thank you. Andrew M. Neiderer US Army Research Lab
On Wed, 30 Nov 2011, Neiderer, Andrew M USA CIV (US) wrote:
With lots of help I think I am getting close to a "hello, world" BGL, MPI example to execute.
Not sure about the error I am getting from execution
I can't tell what you are getting the segmentation fault from. Does a C MPI program work (like the one from http://www.dartmouth.edu/~rc/classes/intro_mpi/hello_world_ex.html) work? If not, see the documentation for your MPI implementation (these types of problems in C MPI code are often due to mismatched software versions). -- Jeremiah Willcock
Thank you. At the URL I simply moved the 2 arguments for main signature, ie int main(int argc, char *argv[]) { ... } and both mpicc and mpiCC generated an executeable for mpirun. But mpiCC obviously does not like int main(argc, argv) int argc; char *argv[]; { ... } ________________________________ From: boost-users-bounces@lists.boost.org [boost-users-bounces@lists.boost.org] on behalf of Jeremiah Willcock [jewillco@osl.iu.edu] Sent: Wednesday, November 30, 2011 9:23 AM To: boost-users@lists.boost.org Subject: Re: [Boost-users] FW: BGL, MPI On Wed, 30 Nov 2011, Neiderer, Andrew M USA CIV (US) wrote:
With lots of help I think I am getting close to a "hello, world" BGL, MPI example to execute.
Not sure about the error I am getting from execution
I can't tell what you are getting the segmentation fault from. Does a C MPI program work (like the one from http://www.dartmouth.edu/~rc/classes/intro_mpi/hello_world_ex.html) work? If not, see the documentation for your MPI implementation (these types of problems in C MPI code are often due to mismatched software versions). -- Jeremiah Willcock
On Wed, 30 Nov 2011, Neiderer, Andrew M USA CIV (US) wrote:
Thank you.
At the URL I simply moved the 2 arguments for main signature, ie
int main(int argc, char *argv[])
{
...
}
and both mpicc and mpiCC generated an executeable for mpirun.
Does that executable run correctly? -- Jeremiah Willcock
yes it did. We have 20 compute nodes; I only did for 8 of them $mpirun -np 8 ./helloWorld Hello world from process 1 of 8 Hello world from process 5 of 8 Hello world from process 0 of 8 ... Hello world from process 7 of 8 ________________________________ From: boost-users-bounces@lists.boost.org [boost-users-bounces@lists.boost.org] on behalf of Jeremiah Willcock [jewillco@osl.iu.edu] Sent: Wednesday, November 30, 2011 1:37 PM To: boost-users@lists.boost.org Subject: Re: [Boost-users] FW: BGL, MPI On Wed, 30 Nov 2011, Neiderer, Andrew M USA CIV (US) wrote:
Thank you.
At the URL I simply moved the 2 arguments for main signature, ie
int main(int argc, char *argv[])
{
...
}
and both mpicc and mpiCC generated an executeable for mpirun.
Does that executable run correctly? -- Jeremiah Willcock
participants (2)
-
Jeremiah Willcock -
Neiderer, Andrew M USA CIV (US)