parallel run with cluster

Dear users
I use cluster for parallel simulation.
I received output file easily before but recently :
output file type one line and don’t print any line after (cpu of cluster are occupied with lammps but lammps dont print any output)

LAMMPS (15 May 2015)
Lattice spacing in x,y,z = 4.086 4.086 4.086
Created orthogonal box = (-408.6 0 0) to (817.2 57.204 49.032)
16 by 1 by 1 MPI processor grid
Created 10080 atoms
Lattice spacing in x,y,z = 5.5 5.5 5.5
Created 5859 atoms
Created 27216 atoms
Lattice spacing in x,y,z = 5 5 5
Created 460 atoms
10080 atoms in group wall
33535 atoms in group flow4
460 atoms in group fluc
33075 atoms in group flow
230 atoms in group flucw
230 atoms in group flucp
dynamic group dflow defined
1260 atoms in group lowers
1260 atoms in group uppers
2520 atoms in group walls
7560 atoms in group walld
Neighbor list info …
2 neighbor list requests
update every 1 steps, delay 1 steps, check yes
master list distance cutoff = 10.52
Setting up run …
Memory usage per processor = 0.439999 Mbytes
Step Time mytemp walldt dflow sourcevo permiatv
0 0 100 100 0 665 -80

what is wrong?

Depends. Is it still running? The output might be buffered so you might not see anything until a large chunk of text is ready to be written.

Dear Paquay
I don’t change my previous script(input file) or number of node(cpu) on cluster
lammps update output file for me about every 2 minutes before but now
after 6 hours lammps don’t update output file

Dear users
I use cluster for parallel simulation.
I received output file easily before but recently :
output file type one line and don't print any line after (cpu of cluster are
occupied with lammps but lammps dont print any output)

it is difficult to say anything specific with so limited information.
chances are, you either are triggering a bug due to having processors
with no atoms in their assigned subdomain or that you have an unstable
simulation.

try compiling the latest development version and check again.
also, try - if possible - running the same input on a different
machine and also possibly interactively and with fewer CPUs.

axel.

Dear Axel
I reduced my CPUs and I printed output each time step (thermo 1) and i saw when i increase number of CPUs(>1 cpu) speed of simulation decrease.
how i can fix that?

Dear Axel
I reduced my CPUs and I printed output each time step (thermo 1) and i saw
when i increase number of CPUs(>1 cpu) speed of simulation decrease.
how i can fix that?

i don't have a crystal ball. how should i know? surely you are doing
something wrong.

axel.

There are mpi bugs sometimes between different system implementations and lammps. I ran an input on one computer last week and one processor in the job got stuck with no error report and lammps just sat there for half an hour. The same input ran fine on another computer but one of the commands was triggering a memory leak. Basically you have to systematically debug what is going on between lammps and your system.