The instructions here assume you know the basics of running NAMD and have done it on a cluster like DT already. The details specific to Stampede2 are presented here. Stampede2 has 1T of space for each user, so clear out your data once in a while.
Your working directory should look similar to this
/work/04431/ewang125/stampede2
Note that you need the stampede2 directory.
Also, you can type cdw
to go to your work directory. To setup your directory for group readability please do the following:
cdw setfacl -d -m u:jklauda:rw .
Equilibrate your system on DT and get your restart files dyn.xsc
, dyn.coor
, dyn.vel
In your namd directory, you need 5 more files to start production:
1. dyn.csh
2. dyn.scr
3. dyn-1.inp
4. dyn-2.inp
5. last.seqno
Example files using the KNL nodes. Use 1 node per job!
prod.tar.gz
Only dyn.csh changes if you're using the SKX nodes. For 30k atoms, 2 nodes is optimal. For 70k atoms, 3 nodes is ok but 4 is not ok. 4 tasks per node is recommended
dyn.csh.gz
An important note: in dyn.scr
, you need to cd directly to the working directory. If you have a lot of jobs, you may find it tedious to change every file.