RES: [Paraview] Common problems with Ensight

Stéphane Montésino Stephane.Montesino at hmg.inpg.fr
Thu Jun 29 13:06:43 EDT 2006


My list is not so short. I simulate turblent flow with a grid 
512x128x256. I've never heard about this problem

I had the same crash when animating in an .avi file with time step 
longer than ...
But when I unchecked the box "cache geometry", it ran really well.

Here is an example of what can be done with this powerful tool on 
windows 2000 and a 128 Mo video card.
ftp://ftp.legi.hmg.inpg.fr/pub/MONTESINO_Stephane/visu/zoom_Eddies_shear_Velocity2.avi 

365Mo

I could signal a problem butthis one depends on the machine over which 
your visualisation is running on.
When the of your graphical object (complex isosurface, glyph of 3d 
field) is bigger than the memory card, it crash on some machine.
On other machine (maybe on SGI), I heard it doesn't crash because the 
machine swap with the RAM.
For more details ask goldmann at idris.fr. He was the one that signal me 
this bug.


Renato N. Elias wrote:

> 
>
>Hi Stephane,
>
>I've tried to save some files opened as "unformatted" following what you
>said some time ago (I remember you've posted something regarding it). The
>problem was that my Fortran compiler (Compaq and Intel 8.x/9.x) was not able
>to save a long list of parameters passed to the WRITE statement (my list of
>parameters is much larger than yours).
>
>Regarding the second point: I know about the "cache geometry". In fact, I
>seldom leave this option enabled when I'm animating something. Actually, my
>question was about the number of time steps. Have you got success animating
>more than 1000 time steps? In my case, PV always crashed before finishing to
>open my case file. It seems PV tries to read and check all files before
>showing the "time step" list on the "select time value" tab.
>
>By the way, I don't know nothing about visualization algorithms but I guess
>PV should have an option in order to allow an offscreen video rendering just
>as Ensight do. Maybe it could improve the video rendering process. Instead
>of loading and rendering each time step, the engine could only record the
>changes between frames. Sorry if I'm saying something stupid... It's just
>speculation from someone that doesn't have any background in visualization
>;-) 
>
>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 
>
>
>-----Mensagem original-----
>De: Stéphane Montésino [mailto:Stephane.Montesino at hmg.inpg.fr] 
>Enviada em: quinta-feira, 29 de junho de 2006 13:18
>Para: paraview at paraview.org; rnelias at nacad.ufrj.br
>Assunto: [Paraview] Common problems with Ensight
>
>Hi Renato,
>
>1)To write your data in an unformatted way,  you should use direct access
>and put the reclength.
>
>      reclength=80*3+4*(1+(imax-imin+1)*(jmax-jmin+1)*(kmax-kmin+1)*ndv)
>
>        open (unit=FileUnit,file=trim(GeoName)//'.geo',&
>              form='UNFORMATTED',access="direct",recl=reclength)
>        write(unit=FileUnit,rec=1)                          &
>     &        binary_form                                   &
>     &       ,file_description1                             &
>     &       ,file_description2                             &
>     &       ,node_id                                       &
>     &       ,element_id                                    &
>     &       ,( part,rank                                   &
>     &         ,description_part                            &
>     &         ,block                                        &
>     &         ,isize(rank),jsize(rank),ksize(rank)            &
>     &         ,(sngl(x1(i)),i=lim(1,1,rank),lim(2,1,rank))    &
>     &         ,(sngl(x2(j)),j=lim(1,2,rank),lim(2,2,rank))    &
>     &         ,(sngl(x3(k)),k=lim(1,3,rank),lim(2,3,rank))    &
>     &         ,rank=1,npart)
>
>
>        open (unit=FileUnit,file=VarFileName,&
>     &        form='UNFORMATTED',access="direct",recl=reclength)
>        write(unit=FileUnit,rec=1) VarFileName           &
>     &                            ,part,rank,block       &
>     &                            ,((((SNGl(var(m,i,j,k))&
>     &                            ,i=imin,imax)          &
>     &                            ,j=jmin,jmax)          &
>     &                            ,k=kmin,kmax)          &
>     &                            ,m=1,ndv)
>
> 7)To animate a lot of time step, uncheck the box cache geometry  in the
>Keyframe animation menu.
>
>Renato N. Elias wrote:
>
>  
>
>>>Guys,
>>>
>>>I'll list some problems related to Ensight's format with ParaView 
>>>that I have observed since I started using ParaView. The intention of 
>>>this post is not to reclaim anything. It's just to help PV beginners 
>>>as well as collect some experiences from other ParaView/Ensight users:
>>>
>>>1) ParaView is not able to open Ensight's binary files written by 
>>>Fortran programs as UNFORMATTED. Thus, if you're a Fortran 
>>>programmer, you
>>>***MUST*** open your files as BINARY (form="binary"), otherwise 
>>>ParaView will not work properly. NOTE: Only some newer Fortran compilers
>>>      
>>>
>(e.g.
>  
>
>>>Intel's Fortran) support binary files;
>>>
>>>2) If you're worried about changing the "byte order" every time 
>>>you're loading an Ensight's binary file and if you're programming in 
>>>Fortran, I'd suggest you open your files with the byte order 
>>>previously defined, e.g.,
>>>open(unit=iens,file='foo.geo',form='binary',CONVERT='BIG_ENDIAN')
>>>
>>>3) There are lots of filters in PV 2.4.3 version that has not been 
>>>working fine (or even disabled) -- a list of the disabled filters can 
>>>be found in previous posts. It seems the PV team has been 
>>>re-developing the Ensight's readers since the PV 2.2.1. Thus, if 
>>>you're looking for a ParaView version more stable with Ensight's 
>>>formats, I guess, you should go through PV 2.2.1;
>>>
>>>4) When working in parallel, the Ensight's server-of-server format 
>>>(SOS) only work to load a dataset and do some basic things. Some 
>>>filters such as "Append datasets", "Clean to grid" and further 
>>>resources like: animate transient datasets are not available yet;
>>>
>>>5) If you're writing a Ensight's SOS file from a program (like me), 
>>>be sure that you're not putting extra blank spaces after the name of the
>>>      
>>>
>case files.
>  
>
>>>I spent a good time comparing almost equal files in order to 
>>>understand what was going wrong. For example:
>>>
>>>machine id: localhost
>>>executable: ensight7.server
>>>directory:
>>>login id:
>>>casefile: "foo_4_00.case                           "
>>>
>>>Is different than
>>>
>>>machine id: localhost
>>>executable: ensight7.server
>>>directory:
>>>login id:
>>>casefile: "foo_4_00.case"
>>>
>>>(***THE QUOTATION MARKS ARE TO REINFORCE WHAT I DID WRONG***)
>>>
>>>6) Lookmark doesn't seem to work with Ensight's format. In all my 
>>>tries PV crashed  :-(
>>>
>>>7) An issue to be asked to the list: Has anyone got success in 
>>>animating more than 1000 time steps with ParaView and Ensight's files?
>>>
>>>
>>>I can't remember nothing more but other issues are welcome  ;-)
>>> 
>>>
>>>Renato N. Elias
>>>      
>>>
>
>
>
>--
>Stephane MONTESINO
>
>PhD Student                     tel:+33 4 76 82 52 91
>LEGI                            fax:+33 4 76 82 70 22
>BP 53
>38041 Grenoble Cedex            email: stephane.montesino at hmg.inpg.fr
>France                          http://www.legi.hmg.inpg.fr 
>
>
>  
>


-- 
Stephane MONTESINO

PhD Student                     tel:+33 4 76 82 52 91
LEGI                            fax:+33 4 76 82 70 22
BP 53
38041 Grenoble Cedex            email: stephane.montesino at hmg.inpg.fr
France                          http://www.legi.hmg.inpg.fr 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Stephane.Montesino.vcf
Type: text/x-vcard
Size: 308 bytes
Desc: not available
Url : http://public.kitware.com/pipermail/paraview/attachments/20060629/60ff64a5/Stephane.Montesino.vcf


More information about the ParaView mailing list