[Paraview-developers] [EXTERNAL] Re: Bad superbuild

Wed Jul 27 16:09:45 EDT 2016

5.1.2 and Superbuilds

Worked like a charm after following Chuck and Utkarsh’s advice.  Thanks team!

From: Paraview-developers [mailto:paraview-developers-bounces at paraview.org] On Behalf Of Scott, W Alan
Sent: Monday, July 25, 2016 7:44 PM
To: Chuck Atkins <chuck.atkins at kitware.com>
Cc: paraview-developers at paraview.org
Subject: Re: [Paraview-developers] [EXTERNAL] Re: Bad superbuild

Mind taking a look at the Wiki, under Build and Install?  The section on ParaView Settings is confusing.  Further, the flags for the MPI_*_COMPILER are a actually different thatn you have here.

http://www.paraview.org/Wiki/ParaView:Build_And_Install#ParaView_Settings

Thanks,

alan

From: Chuck Atkins [mailto:chuck.atkins at kitware.com]
Sent: Monday, July 25, 2016 7:35 PM
To: Scott, W Alan <wascott at sandia.gov<mailto:wascott at sandia.gov>>
Cc: paraview-developers at paraview.org<mailto:paraview-developers at paraview.org>
Subject: Re: [EXTERNAL] Re: [Paraview-developers] Bad superbuild

Yes, that's generally the most reliable way to get the correct and complete MPI support.  I suspect the FindMPI module was getting half-baked information from the wrappers and augmenting with the specified MPI_INCLUDE_PATH.  If you don't want it in the environment then I would try explicitly setting the compiler wrappers and nothing else with:

-DMPI_C_COMPILER:FILEPATH=/opt/openmpi-1.8-gnu//bin/mpicc
-DMPI_CXX_COMPILER:FILEPATH=/opt/openmpi-1.8-gnu//bin/mpicxx

and drop any other MPI_FOO vars, leaving the FindMPI module interrogate the wrappers to populate the rest.

----------
Chuck Atkins
Staff R&D Engineer, Scientific Computing
Kitware, Inc.
(518) 881-1183

On Mon, Jul 25, 2016 at 9:24 PM, Scott, W Alan <wascott at sandia.gov<mailto:wascott at sandia.gov>> wrote:
So, let me ask this.  Should I not point at MPI at all through CMake vars, but just point the PATH including the MPI compiler?

Alan

From: Chuck Atkins [mailto:chuck.atkins at kitware.com<mailto:chuck.atkins at kitware.com>]
Sent: Monday, July 25, 2016 7:20 PM
To: Scott, W Alan <wascott at sandia.gov<mailto:wascott at sandia.gov>>
Cc: paraview-developers at paraview.org<mailto:paraview-developers at paraview.org>
Subject: [EXTERNAL] Re: [Paraview-developers] Bad superbuild

Hi Alan,
The FindMPI module can be sensitive to an enormous number of settings.  Do you have openmpi loaded as an environment module?  If so, I would get rid of any -DMPI_FOO variables from the CMake configure line and just let the FindMPI.cmake module do it's thing.  If the MPI compiler wrappers are in your PATH then it should be able to interrogate them correctly.

----------
Chuck Atkins
Staff R&D Engineer, Scientific Computing
Kitware, Inc.
(518) 881-1183<tel:%28518%29%20881-1183>

On Mon, Jul 25, 2016 at 2:58 PM, Scott, W Alan <wascott at sandia.gov<mailto:wascott at sandia.gov>> wrote:
I am trying to build clusters (again) with the superbuild.  This is Linux, master.  I am using the gnu compiler and openmpi for gnu.  I am having issues with MPI.  I am trying to use a system installed MPI.

Here is what I am sending to cmake:
cMakeVars = -DCMAKE_INSTALL_PREFIX:PATH=/projects/viz/paraview/ParaView/5.1.1-sb-mesa/Linux-chama-x86_64 -DBUILD_TESTING:BOOL=OFF -DCMAKE_BUILD_TYPE:STRING=Debug -DParaView_FROM_GIT:BOOL=OFF -DParaView_FROM_SOURCE_DIR:BOOL=ON -DPARAVIEW_SOURCE_DIR:PATH=/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5 -DPARAVIEW_ALWAYS_SECURE_CONNECTION:BOOL=ON -DENABLE_paraview:BOOL=ON -DENABLE_ffmpeg:BOOL=ON -DENABLE_python:BOOL=ON -DENABLE_numpy:BOOL=ON -DENABLE_png:BOOL=ON -DENABLE_osmesa:BOOL=ON -DENABLE_cgns:BOOL=ON -DENABLE_silo:BOOL=ON -DENABLE_visitbridge:BOOL=ON -DMPI_COMPILER:FILEPATH=/opt/openmpi-1.8-gnu//bin/mpicxx -DPARAVIEW_ALWAYS_SECURE_CONNECTION:BOOL=ON -DENABLE_mpi:BOOL=ON -DPARAVIEW_USE_MPI:BOOL=ON -DMPI_INCLUDE_PATH:PATH=/opt/openmpi-1.8-gnu//include/

If anyone needs my CMakeCache.txt, please ask and I can send it off list.

Here is my failure:

[  4%] Building C object VTK/ThirdParty/tiff/vtktiff/CMakeFiles/vtktiff.dir/tif_thunder.c.o
CMakeFiles/IceTMPI.dir/mpi.c.o: In function `getMPIRequest':
/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5/ThirdParty/IceT/vtkicet/src/communication/mpi.c:110: undefined reference to `ompi_request_null'
/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5/ThirdParty/IceT/vtkicet/src/communication/mpi.c:116: undefined reference to `ompi_request_null'
CMakeFiles/IceTMPI.dir/mpi.c.o: In function `create_request':
/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5/ThirdParty/IceT/vtkicet/src/communication/mpi.c:160: undefined reference to `ompi_request_null'
CMakeFiles/IceTMPI.dir/mpi.c.o: In function `destroy_request':
/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5/ThirdParty/IceT/vtkicet/src/communication/mpi.c:168: undefined reference to `ompi_request_null'
CMakeFiles/IceTMPI.dir/mpi.c.o: In function `icetCreateMPICommunicator':
/projects/viz/paraview/src/ParaView/5.1.1-sb/ParaView5/ThirdParty/IceT/vtkicet/src/communication/mpi.c:200: undefined reference to `ompi_mpi_comm_null'
CMakeFiles/IceTMPI.dir/mpi.c.o: In function `MPISubset':

Thanks all!

Alan

--------------------------------------------------------
W. Alan Scott
ParaView Support Manager

SAIC
Sandia National Laboratories, MS 0822
Org 9326 - Building 880 A1-K
(505) 284-0932<tel:%28505%29%20284-0932>   FAX (505) 284-5619<tel:%28505%29%20284-5619>
---------------------------------------------------------

_______________________________________________
Powered by www.kitware.com<http://www.kitware.com>

Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html

Search the list archives at: http://markmail.org/search/?q=Paraview-developers

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20160727/dd98e4f1/attachment-0001.html>