NAMD is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems. It is written and maintained by the Theoretical and Computational Biophysics Group at the University of Illinois at Urbana–Champaign. More details can be found at http://www.ks.uiuc.edu/Research/namd/.
The easiest way to get started running NAMD on Blue Waters is to run a pre-compiled version of the code. Jim Phillips, lead NAMD developer, maintains a version of NAMD in his home directory which you may use. The latest development version can be found in /u/sciteam/jphillip/NAMD_build.latest/. There are 7 different versions of NAMD to choose from, based on the size of the simulation and whether or not you want to use GPUs in your simulation.
Jim also provides some scripts which will build and submit your jobscript in /u/sciteam/jphillip/NAMD_scripts/runbatch*. Again, there are several versions, differing in which of the 7 different versions of NAMD they use. The "standard" mode would be "/u/sciteam/jphillip/NAMD_scripts/runbatch". Running it with no options gives a short description of how you would submit a simulation run. If desired, you can look at the files it generates to understand how to customize your own job script.
Jim Phillips provides a tool called runbatch that generates and runs a job script for you.
There are 4 variations:
XE CPU run:
XK GPU run:
Memory-optimized XE CPU run
Memory-optimized XK GPU run (memory-optimized input):
Run each command without options to see the list of input parameters
For example, to simulate the system described in mysim.namd using 400 XE nodes, I would run:
When you run the command, it will generate a script, which is output to the screen, and submit that script as a job. To capture the script so that you can customize it, you can redirect the output of runbatch and edit out everything before "#!/bin/tcsh" and everything after the final "aprun ..." line.