<div dir="ltr">I managed to compile it! Thank you.<div><br></div><div>I now need to figure out how to use it...it does not seem as simple as telling the paraview how many cores to use as all I get is visual artifacts instead of real output (I see a field with spots, instead of the expected slice visualization that I get when I only run paraview on a single core). Doing mpirun on pvserver also does not seem to do much (spots, and same speed) while attempting to use pvrenderserver does not work - not without delving into exactly how to make it work - the client connects, but no rendering happens...</div>
<div><br></div><div>I will look at the documentation and the web first before asking for help...</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 18 June 2014 20:46, Victor <span dir="ltr"><<a href="mailto:victor.major@gmail.com" target="_blank">victor.major@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Thank you for the advice. I will attempt this tomorrow.<div><br></div><div>Regarding the openmpi path, I recompiled the system wide openmpi and it indeed lives in /opt/openmpi-1.8.1. I no longer use an alternative openmpi install path as that did not work out anyway.</div>
</div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><br><div class="gmail_quote">On 18 June 2014 20:08, Utkarsh Ayachit <span dir="ltr"><<a href="mailto:utkarsh.ayachit@kitware.com" target="_blank">utkarsh.ayachit@kitware.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Looking back at your earlier email, you say "....where it attempts to<br>
<div>link to the system<br>
> wide openmpi-1.8.1 location and of course fails".<br>
<br>
</div>Looking at the error message, I don't see where it links agains the<br>
system wide openmpi. It's indeed linking against the openmpi in<br>
/opt/openmpi-1.8.1/ which is what you have built, right? Or am I<br>
misreading this?<br>
<br>
You can also manually edit the CMakeCache.txt file and ensure all<br>
"mpi" variables point to directories from your build esp. the<br>
MPI_(C|CXX|Fortran)_* flags.<br>
<span><font color="#888888"><br>
Utkarsh<br>
</font></span><div><br>
On Fri, May 30, 2014 at 2:40 AM, Victor <<a href="mailto:victor.major@gmail.com" target="_blank">victor.major@gmail.com</a>> wrote:<br>
</div><div><div>> Thank you for the fast reply!<br>
><br>
> I compiled an alternative openmpi-1.8.1 with -fPIC flag (I verified that it<br>
> was placed into the Makefile), removed the original Paraview build<br>
> directory, modified the Paraview build configuration by rerunning ccmake to<br>
> point to the alternative openmpi path, but the error remained.<br>
><br>
> During Paraview compilation the alternative path for openmpi-1.8.1 is used<br>
> until it comes to the linking step where it attempts to link to the system<br>
> wide openmpi-1.8.1 location and of course fails.<br>
><br>
> Linking CXX shared library ../../../lib/<a href="http://libvtkParallelMPI-pv4.1.so" target="_blank">libvtkParallelMPI-pv4.1.so</a><br>
> /usr/bin/ld: /opt/openmpi-1.8.1/lib/libmpi_cxx.a(intercepts.o): relocation<br>
> R_X86_64_32S against `ompi_mpi_cxx_comm_throw_excptn_fctn' can not be used<br>
> when making a shared object; recompile with -fPIC<br>
> /opt/openmpi-1.8.1/lib/libmpi_cxx.a: could not read symbols: Bad value<br>
> collect2: ld returned 1 exit status<br>
> make[2]: *** [lib/libvtkParallelMPI-pv4.1.so.1] Error 1<br>
> make[1]: *** [VTK/Parallel/MPI/CMakeFiles/vtkParallelMPI.dir/all] Error 2<br>
> make: *** [all] Error 2<br>
><br>
><br>
> I cannot (will not) overwrite the system wide openmpi as I have a parallel<br>
> job running at the moment (CFD simulation) and this node is acting as a head<br>
> node serving openmpi from an NFS shared directory. I am confident that the<br>
> application will not need to call on openmpi now that it is already running,<br>
> but I'd rather not find out that I am wrong.<br>
><br>
> Thus is there a way to force the Paraview installer to link against an<br>
> alternative build that was compiled with -fPIC?<br>
><br>
> Thanks,<br>
><br>
> Victor<br>
><br>
><br>
> On 29 May 2014 18:37, Utkarsh Ayachit <<a href="mailto:utkarsh.ayachit@kitware.com" target="_blank">utkarsh.ayachit@kitware.com</a>> wrote:<br>
>><br>
>> You'll have to recompile openmpi with -fPIC or link against a shared<br>
>> build of the same.<br>
>><br>
>> On Thu, May 29, 2014 at 5:06 AM, Victor <<a href="mailto:victor.major@gmail.com" target="_blank">victor.major@gmail.com</a>> wrote:<br>
>> > I am trying to compile Paraview 4.1 with MPI support but make exits at<br>
>> > this<br>
>> > point:<br>
>> ><br>
>> > /usr/bin/ld: /opt/openmpi-1.8.1/lib/libmpi.a(comm_init.o): relocation<br>
>> > R_X86_64_32 against `ompi_comm_f_to_c_table' can not be used when making<br>
>> > a<br>
>> > shared object; recompile with -fPIC<br>
>> > /opt/openmpi-1.8.1/lib/libmpi.a: could not read symbols: Bad value<br>
>> > collect2: ld returned 1 exit status<br>
>> > make[2]: *** [lib/libvtkParallelMPI-pv4.1.so.1] Error 1<br>
>> > make[1]: *** [VTK/Parallel/MPI/CMakeFiles/vtkParallelMPI.dir/all] Error<br>
>> > 2<br>
>> > make: *** [all] Error 2<br>
>> ><br>
>> > How can I fix this error?<br>
>> ><br>
>> > Thanks,<br>
>> ><br>
>> > Victor<br>
>> ><br>
>> > _______________________________________________<br>
>> > Powered by <a href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
>> ><br>
>> > Visit other Kitware open-source projects at<br>
>> > <a href="http://www.kitware.com/opensource/opensource.html" target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
>> ><br>
>> > Please keep messages on-topic and check the ParaView Wiki at:<br>
>> > <a href="http://paraview.org/Wiki/ParaView" target="_blank">http://paraview.org/Wiki/ParaView</a><br>
>> ><br>
>> > Follow this link to subscribe/unsubscribe:<br>
>> > <a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
>> ><br>
><br>
><br>
</div></div></blockquote></div><br></div>
</div></div></blockquote></div><br></div>