My first attempt at getting MPI working with paraview was by using the default compiler and setting the MPI_INCLUDE_PATH and MPI_LIBRARY paths. The problem I was running into was that I was getting underfined references when linking the mpi libraries. The solution that Lam recommended was to use their compiler to solve the library problems. If somebody could send me their MPI_INCLUDE_PATH and MPI_LIBRARY configurations it would help me out alot.
<br><br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">David E DeMarle</b> <<a href="mailto:dave.demarle@kitware.com">dave.demarle@kitware.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
That is odd.<br><br>I have a feeling that lam's mpicc brings in or leaves out something important in the compilation and linking, especially when it compiles the non parallel paraview app. When compiling against MPICH for example we don't use MPICH's mpicc. But instead use normal cc and link in the extra libraries. Instead of setting the compiler directly, have you tried just turning on PARAVIEW_USE_MPI and then setting the MPI_INCLUDE_PATH and MPI_LIBRARY instead?
<br><br>cheers<div><span class="e" id="q_113543fa7b4d6324_1"><br>Dave<br><br><br><br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">Robert Maynard</b> <<a href="mailto:thorin.hax@gmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
thorin.hax@gmail.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
the backtrace seems really short to me, I did compile paraview with the Debug flag.<span><br><br>Program received signal SIGFPE, Arithmetic exception.<br></span>0x03d79ac6 in ?? ()<br>(gdb) backtrace<br>#0 0x03d79ac6 in ?? ()
<br>#1 0x00000000 in ?? ()
<br>(gdb) <div><span><br><br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">David E DeMarle</b> <<a href="mailto:dave.demarle@kitware.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
dave.demarle@kitware.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
I was thinking of a stack dump not a system call trace. In gdb after the crash type backtrace and send that output to the list.<br><br>Thanks Robert.<br>Dave<div><span><br><br><div><span class="gmail_quote">
On 6/22/07, <b class="gmail_sendername">
Robert Maynard</b> <<a href="mailto:thorin.hax@gmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">thorin.hax@gmail.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Sure I have attached the strace output<div><span><br><br><br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">David E DeMarle</b> <<a href="mailto:dave.demarle@kitware.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
dave.demarle@kitware.com</a>> wrote:
</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Can you send a stack trace for the floating point exception?<div><span>
<br><br><br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">Robert Maynard</b> <<a href="mailto:thorin.hax@gmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">thorin.hax@gmail.com
</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Yes i understand I have to link to the server in MPI now, but as my first email stated, by I trying running the client now that I compiled mpi support in, I get a runtime floating point exception before paraview launches.
<br>
<br><div><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">David E DeMarle</b> <<a href="mailto:dave.demarle@kitware.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">dave.demarle@kitware.com
</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
In 3.0 the paraview executable is not MPI parallel, only the server executables are.<br>That is, you can no longer run "mpirun -np N paraview" but instead have to do something like "mpirun -np N pvserver&; paraview" and then use File->Choose Server to connect to the parallel server.
<br><br>cheers<br>Dave DeMarle<br><br><div><div><span><span class="gmail_quote">On 6/22/07, <b class="gmail_sendername">Robert Maynard</b> <<a href="mailto:thorin.hax@gmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
thorin.hax@gmail.com</a>> wrote:</span></span></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div><span>
I have compiled paraview with mpi support using lam. I changed the standard c and c++ compilers for paraview too mpicc and mpic++ so that all the<br>mpi references where aviable. The build was correctly compiled, the problem comes when trying to launch the standalone version of paraview.
<br><br>When launching /bin/paraview the error i get is "Floating point exception", using gdb to trace the error I get <br> New process 16350]<br><br> Program received signal SIGFPE, Arithmetic exception.<br> 0x05761ac6 in ?? ()
<br><br>Now if I try to run /bin/paraview with mpirun -np 1, it tells me that paraview did not invoke MPI_INIT before quitting. The problem seems to only <br>happen with standalone version, as /bin/pvserver runs perfectly fine with mpirun
<br>
<br></span></div>_______________________________________________<br>ParaView mailing list<br><a href="mailto:ParaView@paraview.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">ParaView@paraview.org
</a><br><a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.paraview.org/mailman/listinfo/paraview</a><br><br></blockquote></div><br>
</blockquote></div><br>
<br>_______________________________________________<br>ParaView mailing list<br><a href="mailto:ParaView@paraview.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">ParaView@paraview.org</a><br><a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.paraview.org/mailman/listinfo/paraview</a><br><br></blockquote></div><br>
</span></div></blockquote></div><br>
</span></div><br>_______________________________________________<br>ParaView mailing list<br><a href="mailto:ParaView@paraview.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">ParaView@paraview.org
</a><br><a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.paraview.org/mailman/listinfo/paraview</a><br><br><br clear="all"></blockquote></div><br>
</span></div></blockquote></div><br>
</span></div><br>_______________________________________________<br>ParaView mailing list<br><a href="mailto:ParaView@paraview.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">ParaView@paraview.org
</a><br><a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.paraview.org/mailman/listinfo/paraview</a><br><br></blockquote></div><br>
</span></div></blockquote></div><br>