Srun does not appear in the job submission file

New to AiiDA
1

Hello,

I have installed aiida to a local institutional cluster and used the following commands in order:

  1. verdi computer setup --config baseline.yml
    baseline.yml:
label: "localhost"
hostname: "localhost"
transport: "core.local"
scheduler: "core.slurm"
work_dir: "/gpfs/wolf2/cades/phy191/scratch/ksu/aiida"
mpirun_command: "srun -n {tot_num_mpiprocs}"
mpiprocs_per_machine: 128
prepend_text: |
   #SBATCH -p batch
   #SBATCH -A PHY191
   module purge; module load DefApps intel/20.0.4 openmpi/4.0.4  hdf5/1.14.3
   export OMP_NUM_THREADS=1

Also chose the following options from the interactive setup after invoking the command above:

Description []:
Shebang line (first line of each script, starting with #!) [#!/bin/bash]:
Default amount of memory per machine (kB).: 10000000
Escape CLI arguments in double quotes [y/N]: N
  1. verdi -p ks computer configure core.local localhost
Use login shell when executing command [Y/n]: N
Connection cooldown time (s) [0.0]:
  1. verdi code create core.code.installed -n --computer localhost --label pw --default-calc-job-plugin quantumespresso.pw --filepath-executable pw.x

Then I ran the example script from aiida.quantum_espresso plugin:

aiida-quantumespresso calculation launch pw -X pw@localhost -F SSSP/1.1/PBE/efficiency

In the folder where the calculation is submitted, here is the slurm submission script:

#!/bin/bash
#SBATCH --no-requeue
#SBATCH --job-name="aiida-221"
#SBATCH --get-user-env
#SBATCH --output=_scheduler-stdout.txt
#SBATCH --error=_scheduler-stderr.txt
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=128
#SBATCH --time=00:30:00
#SBATCH --mem=9765

#SBATCH -p batch
#SBATCH -A PHY191
module purge; module load DefApps intel/20.0.4 openmpi/4.0.4  hdf5/1.14.3
export OMP_NUM_THREADS=1


'pw.x' '-in' 'aiida.in'  > 'aiida.out'

for some reason, the srun command is not picked up and that causes the calculation to fail. If I add the srun command by hand and submit the calculation by hand, then it runs with no error. Thank you for the help!

Kayahan

2

I think the command line way to run QE does not use MPI version of pw.x. Using the API version of the tutorial in Running a pw.x calculation — aiida-quantumespresso documentation, my problem is resolved.

Thanks,
Kayahan

3

Hi @Kayahan_Saritas

the AiiDA-QE cli sets with_mpi to False per default. You could try running the same command and adding --with-mpi True. I think that this should solve your initial problem.

1 Like