[Paraview] vtk format and mpi paraview

[Berk Geveci]

> Berk and Kent,
> Thanks for the quick response,
>
> The data is unstructured and written one file/processor.
> file001.vtk
> file002.vtk
> ...
> file128.vtk (total processors = 128)
>
> The total x,y,z field contains 128 blocks (or domains).
> 1-Can I put the list of the vtk files into a list_meta.vtk file and load
> only the  list_meta.vtk file?

No. You can't. Legacy files were never designed to be parallel. You
can load them one at a time. You should convert to using XML-based VTK
format or Exodus. Both of those support distributed data like this
well.
There is a actually a reader than can do this with legacy VTK files
but it hasn't been tested in a looong time. If you are interested, I
can give you some pointers on where to start. But after that, you
would be on your own. As far as we are concerned, that functionality
is dead.

> 2-If I use say 32 processors for the visualization, do all 32 processors
> read simultaneously (processor 1 reads file001.vtk, processor 2 reads
> file002.vtk ... etc)? or does one processor read all files?

If you use XML-based VTK or Exodus format, each processor will read
m/n files, where m=number of pieces, n=number of processors. In this
case 128/32.

> I should say that I am using ParaView version 1.6.2.

Ouch. I have no idea if any of the information I provided applies to
1.6.2. I am afraid you are in forgotten land :-)

-berk