I am running QM/MM simulations with Gromacs v2022.3, CP2K v9.1 and PLUMED v2.8.
For restart the simulation, I added “-cpt 2” in the command, which would write the checkpoint file every 2 minutes. But actually, as you can see from the attached figure, the last checkpoint file was generated one hour before the simulation finished (wall time).
What would you recommend for correctly restarting a simulation? Thanks a lot!
How long does one MD step take? Checkpointing can have a delay of up to 2 * nstlist steps. If your QM calculation is very expensive this might take an hour.
But I don’t understand why you did not get a checkpoint file at when mdrun terminated. What is you nstlist value in the md.log log file and at which step number did mdrun stop?
You can restart as normal using the files you have. But nearly an hour of computation will need to be redone.
Thanks a lot for your reply! It takes around 20 seconds on each step, and the nstlist was set as 10.
I also don’t understand it, as I also specified the writing out checkpoint file every 2 minutes, somehow it did not work.
Yeah, I can still restart the simulations, that’s no problem, the problem is that there might be one or two Gaussians (metadynamics simulation) discrepancy. Currently, my solution is that I use a smaller number of steps so that the MD simulation can be finished in the required wall time and a cpt file will be write out at last.
