Hello,
I am trying to save and restart the in.rhodo.cpu file in the GPU directory of LAMMPS. The restart files are being generated but when I am trying to read_restart them, I am getting a segmentation fault. The files are attached.
in.rhodo.cpu_loadRestart (753 Bytes)
in.rhodo.cpu (706 Bytes)
Works fine for me. Are you using the current
version of LAMMPS?
Steve
No, I am using an older version of LAMMPS (of 2013). Were there any changes regarding the restarting mechanism in the new version ? Will update the version and post the result. Thank You
LAMMPS (of 2013). Were there any changes regarding the restarting mechansim
yes, I think so. You can see/search all the changes by date on this page:
http://lammps.sandia.gov/bug.html
Steve
Yeah, It is working with the newer version of LAMMPS. Thank You!.
But the restart values are diverging from the original run by a considerable amount. Is this acceptable? Or should the restart file match in exact with the original run? I am checking only the energy value though other parameters are also diverging.
One probable reason I could think of is that the values at the beginning step of the restart file are differing from the original by a decimal amount and I am guessing that the error is propagating.
If divergence is expected, pardon me for not knowing.
See the doc page for the read_restart command which
discusses exact vs non-exact restarts, and mentions fix
shake explicitly, which is used in the rhodo benchmark.
Steve