[Paraview] trouble building paraview with mpi

[Moreland, Kenneth]

Rank 0 does just as much work as the other ranks.

A compression algorithm called SQUIRT is used to transmit images to the client.  It is a specialized run-length encoding algorithm.  It is described in the paper "Remote Large Data Visualization in the ParaView Framework" that you can get from this link: http://www.cs.unm.edu/~kmorel/documents/EGPGV2006Paper.html

-Ken


On 3/4/09 2:42 PM, "Jim Montine" <jmontine at sgi.com> wrote:

I forgot to ask the last time:

1. Does rank 0 do just as much work as the other ranks when it comes to computation or
does rank 0 worry mostly about coordinating things with the client and the other ranks do
most of the computation.

2.  When remote rendering is enabled, what (if any ) compression is used when transmitting
images to the client.


Thanks

________________________________
From: David E DeMarle [mailto:dave.demarle at kitware.com]
Sent: Wed 3/4/2009 10:46 AM
To: Jim Montine
Cc: paraview at paraview.org
Subject: Re: [Paraview] trouble building paraview with mpi

I think the specifics of configuring that varies dependent on the mpi library.

With some it is as simple as giving the name of the machines in the
right order as arguments on the command line to mpirun, with others it
involves making a hosts file where the first entry in the file is the
hostname that you want to be the rank 0 node and the rest are the
others, and I'm sure others make it much more complicated.

You should be able to test it out with a small mpi hello world from
uname, program, and once you get that working try it out with
pvserver.

On Wed, Mar 4, 2009 at 1:34 PM, Jim Montine <jmontine at sgi.com> wrote:
> You were correct, there was a problem with the MPI library setting.
>
> I have been able to build (and run) pvserver using both openmpi and SGI's
> mpt library.
>
> From what I have read, the client (which in my case is on a PC) communicates
> with the rank0
> process in the mpi group.  What I want to do is have the rank0 thread run on
> the headnode of my
> cluster and have all the other pvserver ranks run on the other compute nodes
> of the cluster.
>
> Is there a recommended way for doing this?
>
>
> Thanks
> ________________________________
> From: David E DeMarle [mailto:dave.demarle at kitware.com]
> Sent: Tue 3/3/2009 9:04 AM
> To: Jim Montine
> Cc: paraview at paraview.org
> Subject: Re: [Paraview] trouble building paraview with mpi
>
> It sounds like cmake wasn't able to tell where the MPI libraries are.
> Open up ccmake, turn on the advanced options, and verify that the MPI
> INCLUDE and LIBRARY and EXTRA_LIBRARY paths are correct.
>
> On Tue, Mar 3, 2009 at 11:43 AM, Jim Montine <jmontine at sgi.com> wrote:
>> I am attemtping to build ParaView-3.4.0.  I follow the instructions on the
>> wiki and I am able to use cmake to configure and generate the makefiles.
>>
>> When I run make, things go along fine until I get here:
>>
>> Linking CXX executable ../../../../bin/DistributedData
>>
>>
>> Then I get a bunch of undefined MPI errors like:
>>
>>
>> /store/jmontine/projects/ParaView-bin/bin/libvtkParallel.a(vtkMPIController.o):
>> In function `vtk
>> MPIController::Finalize(int)':
>> vtkMPIController.cxx:(.text+0x878): undefined reference to `MPI_Finalize'
>>
>> /store/jmontine/projects/ParaView-bin/bin/libvtkParallel.a(vtkMPIController.o):
>> In function `vtk
>> MPIController::Initialize(int*, char***, int)':
>> vtkMPIController.cxx:(.text+0x9fb): undefined reference to `MPI_Init'
>> vtkMPIController.cxx:(.text+0xa1a): undefined reference to
>> `MPI_Get_processor_name'
>>
>> I have tried both hpmpi and sgi's mpi with the same result.  I am using
>> gcc
>> for the compiling.
>>
>> Thanks
>>
>>
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the ParaView Wiki at:
>> http://paraview.org/Wiki/ParaView
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.paraview.org/mailman/listinfo/paraview
>>
>>
>
>
>
> --
> David E DeMarle
> Kitware, Inc.
> R&D Engineer
> 28 Corporate Drive
> Clifton Park, NY 12065-8662
> Phone: 518-371-3971 x109
>



--
David E DeMarle
Kitware, Inc.
R&D Engineer
28 Corporate Drive
Clifton Park, NY 12065-8662
Phone: 518-371-3971 x109



   ****      Kenneth Moreland
    ***      Sandia National Laboratories
***********
*** *** ***  email: kmorel at sandia.gov
**  ***  **  phone: (505) 844-8919
    ***      web:   http://www.cs.unm.edu/~kmorel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20090304/c153068c/attachment.htm>