Running CDFT Tutorial Calculation on Cluster

Brian Day 22b... at gmail.com
Mon Jul 30 20:00:41 UTC 2018


Hi Nico,

I use the following intel compilers: module load intel/2017.3.196 
intel-mpi/2017.3.196 cp2k/6.1

Summarized below are the conditions I used and the results I got:
nodes = 1, tasks = 14, executable = cp2k.popt -i *.inp -o *.out ---> Cube 
files generated without issue
nodes = 1, tasks = 14, executable = mpirun -np $SLURM_NTASKS cp2k.popt -i 
*.inp -o *.out ---> 

forrtl: severe (174): SIGSEGV, segmentation fault occurred

Image              PC                Routine            Line        Source

cp2k.popt          000000000D730E14  Unknown               Unknown  Unknown

libpthread-2.17.s  00002ABEB44735E0  Unknown               Unknown  Unknown

libc-2.17.so   00002ABEB61E74DC  cfree                 Unknown  Unknown

cp2k.popt          000000000D767FA8  Unknown               Unknown  Unknown

cp2k.popt          000000000109F105  qs_dispersion_typ         135  
qs_dispersion_types.F

cp2k.popt          0000000000B88BFF  qs_environment_ty        1476  
qs_environment_types.F

cp2k.popt          0000000000B13020  force_env_types_m         232  
force_env_types.F

cp2k.popt          0000000000EE66E0  f77_interface_mp_         335  
f77_interface.F

cp2k.popt          000000000043BEF8  cp2k_runs_mp_run_         405  
cp2k_runs.F

cp2k.popt          0000000000432814  MAIN__                    281  cp2k.F

cp2k.popt          000000000043151E  Unknown               Unknown  Unknown

libc-2.17.so   00002ABEB6188C05  __libc_start_main     Unknown  Unknown

cp2k.popt          0000000000431429  Unknown               Unknown  Unknown
Note that I had to run these simulations on a different cluster here, as 
the one I was using previously requires the submission file to declare a 
minimum of 2 nodes. I can talk to our computing center and see if they can 
test this themselves. 

To (hopefully) clarify my earlier message, when I ran with 2 nodes, the 
last line in the cp2k output file would be:

 The sum of alpha and beta density is written in cube file format to the 
file:


 
/scratch/slurm-1244788/water-dimer-frag-b-pbe-energy-ELECTRON_DENSITY-1_0.cube
and the electron density file would appear in the submission directory, but 
it would be empty. If I re-ran the simulation and passed this blank file to 
the cluster which I was running on, it would run successfully, but when 
trying to open the file in another program such as Avogadro, or using it in 
a subsequent simulation, it would not work. Maybe this is because it is 
only partially writing as you pointed out. 

I will try and update this with the compiling information and a detailed 
debugging output. (Sorry if any of the above does not make sense, I am 
still fairly new to computational work). 

Thanks again.

-Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180730/4091d2e4/attachment.htm>


More information about the CP2K-user mailing list