Dear Steve, Lammps users,
I am having a interesting problem with thermo_style custom command. Basically, I am printing out about 66 quantities (64 of them is from fix ave/spatial vector) in log file using thermo_style custom command. Whenever, I use more than two processors, the simulations are crashing with message “rank 0 in job 109 cluster_PID caused collective abort of all ranks”. I have about 60,000 atoms in my simulation. It runs fine on two processors but is taking a hell lot of time as one can imagine. When I commented the thermo_style command, the simulations are running fine even on 8 processors. How can I tackle this issue? Would flush command of some sort help?
Any insights appreciated.
Regards,
Vikas