I’ve recently compiled the latest version of LAMMPS on our power5 cluster running suse linux. While running the protein example, the job suddenly dies with a segmentation fault, that seems to be reproducible. Attached you’ll find the makefile and queue submit file. The thermodynamic properties printed in the output file look normal, with the only error being the segmentation fault. Restarting from the crashed position produces the same error later on. I know it’s a memory issue, but don’t know the cause. Any insight would be appreciated.
Makefile.power5 (1.52 KB)
sub_poe (1.25 KB)