Dear Stephane,
I have some problems when I try to export as gnuplot format large files.
* I open a large file (644 Mb, maximum level = 16) with the interactive
viewer and a gfv file (option res = 1 in the "View" section). This gfv file
shows the full result (zoomed out), with only isolines of T. Then I go to
"File -> Save", write the name "test.gnu", and then press "Save". The viewer
immediately closes. I get the following message on the terminal:
"
[5] 5434
[4] Segmentation fault gfsview2D file.gfs t.gfv
"
* The problem is not limited to one file, since it happens for different
simulation files.
* I am able to save as ppm image from the same files (gfs and gfv).
* The computer I am using has 23.4 Gb of memory. I don't know if memory can
be an issue here.
* When the file is open in the viewer, I go to "View -> Preferences ->
Advanced -> Resolution", I put 0 instead of 1, and when I press "Enter", the
viewer stops in a similar way.
* If I change the gfv file to "res = 0", the viewer opens and immediatly
closes with the same error: "[2]+ Segmentation fault gfsview2D
file.gfs t.gfv"
* If I zoom in, I am able to save smaller regions of the same file. The
largest file I was able to save was 3.2 Mb, containing 109296 lines. Then
expending the view (zooming slightly out) results in the same error (closing
the viewer and segmentation fault error message).
* I installed a previous version of the viewer (gfsview-snapshot-110122). I
am able to save successfully the gnuplot file from the previous version with
the same files (gfs and gfv). I am able to switch to the option "res = 0"
without crashing the viewer (only very slow). It is also possible to save
the gnuplot file with the "res = 0" option (gnuplot file is 7.3 Mb in the
old format, 368688 lines).
* All the tests above are working fine when looking at smaller files with
similar parameters with the new version of the viewer.
Therefore, it is a problem with this new version of the viewer (after
gfsview-snapshot-110122), related to large files.
Thank you for your help. Please tell me if you need more information, or if
you want one large file to reproduce the error.
Best regards,
Marie-Jean.
2011/9/7 Stephane Popinet <s.popi...@niwa.co.nz>
> > * I have tried also to do the installation without "--disable-mpi", but
> > instead of "$ sudo make install" at the end, I have done "$ make
> install".
> > The installation also finished in a similar way as with the option
> > "--disable-mpi" (result at the end). I don't know why the "sudo" makes a
> > difference.
> Neither do I, it must be something peculiar to the way your system is
> configured.
> > * However, when I export in gnuplot format, the gnuplot files are
> different.
> > ...
> > The size of the new gnuplot file is approximately half the size of the
> > previous version I had.
> Yes, this is due to this patch:
> 2011-05-27 Stephane Popinet <popinet at users.sf.net>
> * Isolines are now continuous (in 2D only)
> > * However, generating the gnuplot files from the batch viewer (with the
> same
> > gfv file) gives different results: the resolution is much lower. The
> script
> > I am using is the following:
> > (for k in `seq 0 1 9`; do
> > if [ -e "file$k.gfs" ]; then
> > cat file$k.gfs
> > echo "Save file$k.gnu { format = Gnuplot }"
> > fi
> > done ) | gfsview-batch2D t.gfv
> > Should it give the same result as with the interactive viewer? This would
> be
> > very useful for me.
> Yes it should. I will look at it.
> cheers
> Stephane
> --------------------------------------------------------------------------- ---
> Malware Security Report: Protecting Your Business, Customers, and the
> Bottom Line. Protect your business and customers by understanding the
> threat from malware and how it can impact your online business.
> http://www.accelacomm.com/jaw/sfnl/114/51427462/
> _______________________________________________
> Gfs-users mailing list
> Gfs-us...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/gfs-users
--
Marie-Jean Thoraval
PhD student, Physical Sciences and Engineering
King Abdullah University of Science and Technology
Jeddah, Saudi Arabia
+966 2 808 21 61
+966 5 67 79 08 33
http://www.carva.org/marie-jean.thoraval