[lammps-users] running lammps in parallel

Hi ,
I have problem with running lammps in parallel using mpi. I am using aeribo potential to simulate cnt bundles. When I have a system consists of 15000 atoms, I am able to run the calculation on multi processors. However, when the atom number increases to 47000, the calculation has no problem when I just use single processor, but gives me the following error when using multi-processor:

p2_16182: p4_error: interrupt SIGSEGV: 11
rm_l_2_16183: (0.210938) net_send: could not write to fd=5, errno = 32
p1_16176: p4_error: interrupt SIGx: 13
p3_16188: p4_error: interrupt SIGx: 13
p0_16170: p4_error: interrupt SIGx: 13
p0_16170: (6.378906) net_send: could not write to fd=4, errno = 32

I would appreciate if anyone can help me with this issue.



I know of no problem with running systems of that size in parallel (or serial).
There is not enough info in your message to provide a better guess. The
error looks like a lo-level message passing issue.