[Paraview] Re: MPI makes Paraview slower?

[Nancy]

>------------------------------
>
>Message: 3
>Date: Tue, 09 May 2006 10:19:30 +0200
>From: Christoph Moder <mail at christoph-moder.de>
>Subject: Re: [Paraview] Re: MPI makes Paraview slower?
>To: paraview at paraview.org
>Message-ID: <44605092.9080104 at christoph-moder.de>
>Content-Type: text/plain; charset=UTF-8; format=flowed
>
>Nancy wrote:
>> Bryan,
>> Slower or faster is decided by the data type. Only
>> regular types of data (structured grid or image data)
>> can be divided into sub-grids automatically by paraview
>> when running in parallel mode.
>> For unstructured data (poly data or unstructured grid)
>> the distribution should be done by users. Or, you can use
>> D3 filter on your data, which can redistribute your data
>> on the parallel system evenly.
>
>Could you describe how to use the D3 filter? I guess you use a Tcl 
>script, but I don't know how to do that.

No, no need use Tcl script, just load the data (e.g. blow.vtk), and choose
the D3 from Filter menu. If you set the Composite option to zero in 3D View Properties,
you will find that before D3 the every process renders the whole blow, but after
D3 every process only render 1/4 blow.

Nancy

>
>Thanks,
>Christoph
>