My CONTCAR file in vasp is coming out to be 0kb, can someome please guide how what I am doing wrong??
my script
#PBS -N 0ps
#PBS -l nodes=1:ppn=24
#PBS -l walltime=30000:00:00
#PBS -q batch
#PBS -V
#PBS -S /bin/bash
EXEC=/opt/software/vasp/vasp.5.4.1/bin/vasp_std
###suanli=em.tpr###
### Set intel environment###
source /opt/intel/composer_xe_2015/bin/compilervars.sh intel64
source /opt/intel/mkl/bin/intel64/mklvars_intel64.sh
source /opt/intel/impi/5.0.2.044/bin64/mpivars.sh
cd $PBS_O_WORKDIR
NP=`cat $PBS_NODEFILE | wc -l`
NN=`cat $PBS_NODEFILE | sort | uniq | tee /tmp/nodes.$$ | wc -l`
cat $PBS_NODEFILE > /tmp/nodefile.$$
mpirun -genv I_MPI_DEVICE ssm -machinefile /tmp/nodefile.$$ -n $NP $EXEC -s -deffnm em -v
rm -rf /tmp/nodefile.$$
rm -rf /tmp/nodes.$$
Hi, job script only cannot help know where the issue comes from.
One comment only for the job: don't use such large wall time "30000:00:00". You'll never expect a VASP job running such long time.
Several to several hundred hours are sufficient for general DFT in modern HPC architectures.
Are you getting the other files, like WAVECAR, CHGCAR, etc., in your local directory? I am wondering if some of these files are getting written to /tmp/nodefile.$$ and then getting removed a the end of the job. Just speculation.
The program will write the data in CONTCAR file after each ionic step if SCF cycle is converged. Maybe you would check whether SCF procedure can converge during the first ionic step?
If it is ok, I would suggest to give some time for that the program can write all necessary files after calculation is done (maybe with the keyword "sleep" in your slurm script). Sometimes, my calculations is finished successfully, but I couldn't have WAVECAR file. But I have never happened it with CONTCAR file.
Regards.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence