[Paraview] Serial & Parallel - Scientific Data

[Utkarsh Ayachit]

> This is not happening for me because all my file are simple legacy .VTK files. Right?
> XML formatted files have this capacity even though they are not .P + (VTS, VTI, VTR)

For structured datasets, yes. Note, there's still a tradeoff. With the
p* versions where you're presplitting the files, all ranks need not
open the same files and as a result will encounter less contention of
parallel file systems, than if you use a the non-p versions which save
everything in a single file. Whether the work of splitting is worth it
depends on how the data was generated in the first place and then how
many ranks/nodes you're planning to run this on. So I'd recommend
doing some testing on your target platform and then adopt one or the
other approach.

> 1. How to transform my simple legacy VTK files in XML file format? Only rewriting them from the begginning?

You can simply load it in ParaView or VTK and save it out as one the XML files.

> 2. Any thoughts on XML written files versus .PVTK generated by ParaView when saving data?

I am not sure what you mean. Can you elaborate? If you're referring to
tradeoff between .p* and non-p versions of the XML files then see the
note earlier.

Utkarsh