program
mkrunfile.py [options] program p1=v11,v12,... p2=...
runfiles are useful if your code is single-core but you still want to make use of modern multi-core systems, or how HPC systems set up their virtual CPUs.
-r rundir add a rundir -l logfile add a logfile
% mkrunfile.py hackcode1 eps=0.025,0.05
tol=0.5,1 seed=1 > test.run
% cat test.run
hackcode1 rundir=run_0001 eps=0.025 tol=0.5 seed=1
hackcode1 rundir=run_0002 eps=0.05 tol=0.5 seed=1
hackcode1 rundir=run_0003 eps=0.025 tol=1 seed=1
hackcode1 rundir=run_0004 eps=0.05 tol=1 seed=1
after this there are currently three ways to process these runfiles:
# simple bash
bash test.run
# gnu parallel
parallel -j 4 < test.run
# nemo’s sbatch frontend
sbatch_nemo.sh test.run
4-feb-2025 drafted, many features no implemented PJT