#!/bin/bash
#.Submit a GPU job to PPPL's Portal cluster.
#SBATCH --gres=gpu:V100:1 #.Number of V100 GPUs requested (max: 2).
#SBATCH -t 48:00:00 #.Time of simulation (max 48 hours).
#SBATCH -n 1 #.Number of nodes (only 1 GPU node, so should be 1).
#SBATCH -J job-name #.Job name.
#SBATCH -o job-name-%j.out #.Name of stdout output file.
#SBATCH -e job-name-%j.err #.Name of stderr error file.
#SBATCH -p centos7 #.Operating system (MUST BE centos7).
#SBATCH --mem-per-cpu=16000 #.Memory per core (Portal V100 GPU has max 16 GBs).
#.Launch MPI code.
#.Note that aliases may not work in compute nodes. One must
#.provide the full path to the gkyl executable.
/location/of/gkyl/executable/gkyl name-of-input-file.lua