[FLASH-USERS] LaserSlab Segmentation Fault

Descamps, Adrien adescamp at slac.stanford.edu
Wed Jul 1 21:12:25 EDT 2020


Dear all,

I am running into segmentation fault when running the LaserSlab example in parallel. I already posted this problem a week ago and I conducted more tests to try to figure out where the issue might be coming from. My initial guess is that the segmentation might be happening when writing part the hdf5 files.
Please find below details of the tests.

  *   I tried to build the LaserSlab problem without the flag +hdf5typeio. It runs fine using only one processor and I have no issue writing the files. As a matter of fact it is running right now. When using 2 processors, the calculations stops at step 78 with a segmentation fault (see lasslab_nohdf5typeioflag_parallel.log).
  *   When building the LaserSlab probem with the flag +hdf5typeio. It crashes at the same point as the case without the flag (see lasslab_hdf5typeioflag_parallel.log).
  *   Building the LaserSlab probem with the flag +hdf5typeio, I modified the number steps to write the plot file to 10 steps and the checkpoint file to 20 steps. The code ran up to step 60 and wrote the plot file and checkpoint file up to this step. (see lasslab_hdf5typeioflag_parallel_shorterWritingSteps.log)

I am using openmpi and 2 processors with 8 GB of memory. If I remove the -fopenmp flag for hypre I get similar results. Meaning that the calculation crashes at step 60 for the same parameters as bullet point number 3.
I am enclosing the submission file and the Makefile.h used to build all the examples above.

Thank you very much for your help,
Adrien


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lasslab_nohdf5typeioflag_parallel.log
Type: application/octet-stream
Size: 86788 bytes
Desc: lasslab_nohdf5typeioflag_parallel.log
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lasslab_hdf5typeioflag_parallel.log
Type: application/octet-stream
Size: 86940 bytes
Desc: lasslab_hdf5typeioflag_parallel.log
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lasslab_hdf5typeioflag_parallel_shorterWritingSteps.log
Type: application/octet-stream
Size: 85306 bytes
Desc: lasslab_hdf5typeioflag_parallel_shorterWritingSteps.log
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: submit.sh
Type: application/x-sh
Size: 319 bytes
Desc: submit.sh
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment.sh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile.h
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20200702/b2cafb07/attachment.h>


More information about the flash-users mailing list