Dear all,
I am interested in running benchmark tests in our latest supercomputer, CRAY XC40 for LAMMPS for scalability studies and parallel efficiency. I can see there are two types of input files, one fixed and one is scaled. Can anyone please tell me how exactly to use them and how do their use differs. How does LAMMPS treat them differently and give parallel efficiency plots as shown in Benchmarks section of LAMMPS webpage.
I will be really grateful if I am giving anything related material that helps me gain more insight into the matter.
P.S.: I am mostly working on parallelism on CPUs and not much encouraged to see parallelism available for LAMMPS on GPUs yet.