Slurm job post-processing delays due to epoch2d.dat and normal.visit updates

[DrakeWhu]

Hi again,

I've been working with EPOCH to run a parameter scan on a LWFA experiment. I've been writing some Python code to interact with EPOCH and Slurm, as I am using a calculation cluster with 3 nodes. The thing here is that when the simulation is clearly finished (all the .sdf files have been generated), the slurm job takes easily 5 more minutes to finish. I realized it is because the updating of epoch2d.dat and normal.visit is taking longer than the simulation to finish.

This problem is not really a big one, but if I wanted to optimize the parameters of Slurm to make the simulations faster, it wouldn't matter if it's gonna take the same amount of time in actualizing the rest of the files.

This is a problem for me because I am working with a simulation database where I am saving the times when the simulation starts and finishes. I'd like the finishing time to be the one from the simulation and not the updating of the files, that way I could do something like a bayesian optimization of the slurm parameters to make it as fast as I can.

As I am not using the files that take longer to update (whenever all the dumps have been made, I just use sdf_helper to save them into a npy file and then from there I do the data analysis), I'd like to know if maybe there are some compiling flags that make this possible. If the compiling flags used for the installation are needed, I can specify them too

Thanks,
Juan