[lammps-users] building lammps for cluster


I had a few general questions about building LAMMPS for a large cluster. The cluster I am working with is:

- IBM Linux Cluster 1350
- 512 2 way IBM X335 nodes where each node contains 2 Xeon P4 processors
- Myrinet interconnect

Does anyone have any recommendations of compiler flags to use for optimization, etc... I have been successful in building the code on the cluster but am really just looking for tips on making it run quickly/efficiently.



The only thing to really care about is single-processor performance,
so you can run the benchmarks and play with compiler settings.

There isn't typically much you can do for parallel settings.