[Paraview] RES: [vtkusers] Extracting an unstructured grid from
Ensight case files...
Renato N. Elias
rnelias at nacad.ufrj.br
Wed Mar 29 17:44:14 EST 2006
I'm afraid of trying any other file format since the problem is mainly due
the Fortran. I really like the Ensight format, it seems me to be the easiest
and more powerful format among the formats supported by PV, but these
weirdness regarding the Fortran binary are really annoying...
I know that the "unformatted" data written by Fortran is something weird for
any other programming language, but what about the "binary" format provided
by some compilers?! What's wrong with this BINARY? My files are all open as
"BINARY" and not as "UNFORMATTED". I use the Compaq and Intel compilers and
both have support to "BINARY". PV (2.2.1) is able to open the files but only
in serial mode. When I try to use a SOS file to open several pieces of model
in parallel they don't work. Can anyone point me the differences between
this kind of "BINARY" (not UNFORMATTED) and the binary provided by other
languages, such as C/C++? I think this "BINARY" should work as in any other
language.
Renato.
________________________________
De: Randall Hand [mailto:randall.hand at gmail.com]
Enviada em: quarta-feira, 29 de março de 2006 17:22
Para: Renato N. Elias
Cc: paraview at paraview.org
Assunto: Re: [vtkusers] Extracting an unstructured grid from Ensight case
files...
Well, i'm about as far away from an Ensight guru as you can get, so someone
else will have to chime in on the specifics of the File Format & Ensight.
But there's a few issues with Binary & Fortran files.
1) Binary isn't really "cross platform compatible" . If you do an
unformatted write on a big endian machine, you have to be careful to byte
swap it on a little-endian machine. This can't be easily autodetected, so
alot of the readers have the ability to specify it. Maybe this option is
missing?
2) Fortran insists on writing Record Markers at the beginning & end of
each write. Handy for integrity checking, royal PITA when you start trying
to use it in anything other than fortran. Again, hard to autodetect, so
some of the readers give you the ability to specify C or Fortran (like the
Plot3d Reader).
So you're kinda left to using other formats. Personally, I would like to
see the new VTK Parallel XML formats extended to allow for putting the data
& the grid in separate files, but I doubt that will happen anytime soon.
Xmdf & Xdmf are an option, but not very ensight friendly (I think).
Personally i've started writing things to VTK XML formats with the
compression enabled. It's ascii, and the grid gets carried around, but it's
a minor annoyance for the stuff I'm doing. The ones where it is a huge
annoyance, we're using PLOT3D files so it's not an issue.
On 3/29/06, Renato N. Elias <rnelias at nacad.ufrj.br> wrote:
Hi Randall, thanks for your advice...
This subject has been drive me crazy!!!
- I can't read SOS files since they are written by Fortran routines;
- I can't save transient data in VTK since VTK does not have support
to
transient data... ok, it can be overcame through PVD files;
- I can't save grid and results separetely... Ok, I can save them
together
but it would give unnecessary data being replicated;
- I tried the Stephane approach (opening the file as unstructured
and saving
with only one write) but the number of parameters passed to the
"write"
function is greater than the limit that the function supports to
write;
- I open the file (ENSIGHT6) as binary (my Fortran has support to
it) but PV
only read my files in serial modes (SOS does not work)
- Ensight Gold binary written by Fortran does not work as well
- Ok, Ensight format in ASCII works fine but it would give me huge
and very
slow files...
- Ok, there's a Fortran/VTK library distributed with PV, but I'd
have to
install VTK or port the library to any machine I use (and I'm not a
system
administrator to do it).
- I only get success when reading Ensight6 Fortran/Binary data in
serial
mode.
I'm feeling that I (and other Fortran programmers) have no choice...
I know
that PV and VTK have different Ensight readers, but why do the VTK
readers
work with my files and in PV it is not always possible? Furthermore,
how can
Ensight read Fortran unstructured/binary and C/C++ files without any
problem? Within the files have a "C binary" and a "Fortran binary"
indication, can't it be used to guide the right way of reading the
file?
I don't know what I should do... I was writing this script to
convert a
bunch of results that I have stored but I'm almost giving up and
going
through write a C writer for my Fortran programs (unfortunately, it
won't
solve the problem of my results stored - only if I do another file
converter)
Regards
Renato N. Elias
===============================================
PhD student - http://www.nacad.ufrj.br/~rnelias
High Performance Computing Center
Federal University of Rio de Janeiro
Rio de Janeiro, Brazil
+55(21) 2562-8080
________________________________
De: Randall Hand [mailto: randall.hand at gmail.com]
Enviada em: quarta-feira, 29 de março de 2006 16:23
Para: Renato N. Elias
Cc: vtkusers at vtk.org; paraview at paraview.org
Assunto: Re: [vtkusers] Extracting an unstructured grid from Ensight
case
files...
Nice to see someone else using Tcl/Tk for file conversion :) I do
it all
the time here.
Sadly, you've run into the #1 problem with VTK formats, the Grid &
Data must
remain together ( AFAIK, someone please correct me if I'm wrong)..
If you
want to separate the two, then you have little choice than to switch
to
another format.
On 3/29/06, Renato N. Elias <rnelias at nacad.ufrj.br> wrote:
Hi there,
Since I'm a poor Fortran programmer that can't make the
Ensight's
binary
format working in Paraview properly, I've been writing a
TCL/VTK
script to
convert from Ensight case files to Paraview/VTK format (and
I hope
it will
work out my problems...).
A have the following situation to cover:
1- Model partitioned in several case files (sometimes I run
problems
in
parallel/MPI)
2- Each case file has its geometry and results files
associated
3- The geometry does not change in time
Then, I was thinking about do the following:
1- Read each case file
2- Export the Unstructured grid through the
vtkXMLUnstructuredGridWriter
3- Export the results for each time step through the
vtkXMLDataSetWriter
4- Assembly the files through a PVD file (Paraview format)
as is
done with
the case file
My problem is:
* * * How do I export unstructured grids and results
separetely? * *
*
I'm using the following script, but it's only able to save
grid+results in
the same file. Note that it's not good since I'll be
replicating
grid that
does not change in time.
--------------------------------------------------------------
package require vtk
# loading command line arguments
set arch [lindex $argv 0]
set nprocs [lindex $argv 1]
# building the case file names
append arch "_"
append arch $nprocs
append arch "_"
# loop over the case files that will be read and saved in
vtk format
for {set i 0} {$i < $nprocs} {incr i} {
set filename $arch
if {$i < 10} { append filename "0" }
append filename $i
set casefile $filename
append casefile ".case"
puts " Reading Ensight casefile $casefile"
vtkGenericEnSightReader reader
reader SetCaseFileName $casefile
reader Update
vtkXMLUnstructuredGridWriter vtkfile
vtkfile SetInput [ reader GetOutput ]
vtkfile SetFileName "$filename.vtk"
puts " Exporting VTK unified file..."
vtkfile Write
reader Delete
vtkfile Delete
}
wm withdraw .
Exit
--------------------------------------------------------------
Thanks for any help
Renato N. Elias
===============================================
PhD student - http://www.nacad.ufrj.br/~rnelias
<http://www.nacad.ufrj.br/~rnelias>
High Performance Computing Center
Federal University of Rio de Janeiro
Rio de Janeiro, Brazil
+55(21) 2562-8080
_______________________________________________
This is the private VTK discussion list.
Please keep messages on-topic. Check the FAQ at:
http://www.vtk.org/Wiki/VTK_FAQ
Follow this link to subscribe/unsubscribe:
http://www.vtk.org/mailman/listinfo/vtkusers
--
Randall Hand
Visualization Scientist,
ERDC-MSRC Vicksburg, MS
Homepage: http://www.yeraze.com <http://www.yeraze.com>
--
Randall Hand
Visualization Scientist,
ERDC-MSRC Vicksburg, MS
Homepage: http://www.yeraze.com
More information about the ParaView
mailing list