Run this example as:

mpirun -np 4 lmp_linux -partition 4x1 -in in.prd

mpirun -np 4 lmp_g++ -partition 4x1 -in in.prd

mpirun -np 8 lmp_g++ -partition 4x2 -in in.prd

You should be able to use any number of replicas >= 3.

LAMMPS (15 Feb 2016)

Running on 4 partitions of processors

Step CPU Clock Event Correlated Coincident Replica

100 0.000 0 0 0 0 0

100 0.539 0 0 0 0 0

100 0.694 0 0 0 0 0

100 0.850 0 0 0 0 0

100 0.928 0 0 0 0 0

200 1.555 400 1 0 4 1

300 1.924 500 2 1 1 1

500 3.495 772 3 0 2 2

800 5.446 1328 4 0 2 2

1000 7.037 1636 5 0 1 3

1000 7.989 1636 5 0 1 3

1200 8.838 1908 6 0 1 2

1300 9.212 2008 7 1 1 2

1900 12.291 4024 8 0 1 3

1900 12.832 4024 8 0 1 3

1900 13.157 4024 8 0 1 3

1900 13.320 4024 8 0 1 3

2100 14.090 4220 9 0 2 3

Loop time of 14.0941 on 4 procs for 2000 steps with 511 atoms

LAMMPS (28 Sep 2016)

Running on 4 partitions of processors

Step CPU Clock Event Correlated Coincident Replica

100 0.000 0 0 0 0 0

100 0.521 0 0 0 0 0

100 0.670 0 0 0 0 0

100 0.822 0 0 0 0 0

100 0.896 0 0 0 0 0

200 1.555 400 1 0 4 1

300 1.918 500 2 1 1 1

500 3.476 772 3 0 2 2

800 5.379 1328 4 0 2 2

1000 6.914 1636 5 0 1 3

1000 7.859 1636 5 0 1 3

1200 8.658 1908 6 0 1 2

1300 9.018 2008 7 1 1 2

1900 12.005 4024 8 0 1 3

1900 12.539 4024 8 0 1 3

1900 12.861 4024 8 0 1 3

1900 13.027 4024 8 0 1 3

2100 13.798 4220 9 0 2 3

Loop time of 13.8021 on 4 procs for 2000 steps with 511 atoms

LAMMPS (28 Sep 2016)

Running on 4 partitions of processors

Step CPU Clock Event Correlated Coincident Replica

100 0.000 0 0 0 0 0

100 0.289 0 0 0 0 0

100 0.373 0 0 0 0 0

100 0.458 0 0 0 0 0

100 0.500 0 0 0 0 0

200 0.800 400 1 0 4 1

300 0.999 500 2 1 1 1

500 1.834 772 3 0 2 2

800 2.864 1328 4 0 2 2

1000 3.706 1636 5 0 1 3

1000 4.211 1636 5 0 1 3

1200 4.655 1908 6 0 1 2

1300 4.854 2008 7 1 1 2

1900 6.503 4024 8 0 1 3

1900 6.792 4024 8 0 1 3

1900 6.968 4024 8 0 1 3

1900 7.055 4024 8 0 1 3

2100 7.463 4220 9 0 2 3

Loop time of 7.46531 on 8 procs for 2000 steps with 511 atoms