From: John Stone (
Date: Tue Sep 16 2008 - 13:35:03 CDT

  Not all of the VMD code is multithreaded yet, so the CPU utilization
you see will depend greatly on what kind of calculations you're doing.
For things that are heavy in Tcl/Python scripting, you'll only be using
one CPU most of the time except when running VMD commands that are implemented
using multiple processors. Most of the "easy" stuff in VMD is not
multithreaded yet. At present the items that make most use of multiple
processors are things like electrostatic potential calculations,
"within" selections, bond determination, and related computations that
scale in complexity with the size of the structure. We're working on doing
the same for trajectory analysis routines, but much remains to be done yet.
VMD renderings with Tachyon (or TachyonInternal) will use all CPUs.

If you can tell me what your batch scripts are doing, I can comment more
on what CPU utilization you should expect to see.

  John Stone

On Tue, Sep 16, 2008 at 02:47:16PM +0200, Benjamin Stauch wrote:
> dear community,
> i'm running VMD in batch mode on a PBS cluster qsub'ing a job array script.
> when i start VMD (on a single machin), it says "multithreading
> available" and correctly prints out the number of processors i have. in
> my job array script, i request a single node and 8 processors per job.
> in the logfile produced by PBS on completion of the job, it puts that
> the job is executed on all 8 processors. but when i log in to a
> particular node (while running my job) and issue a 'top', i only get
> 100% CPU load (which should be ~800%). i also have slight doubts because
> of the speed of my computation (i would expect it to be faster if all 8
> processors were used).
> is there a way to find out if all 8 processors per node are used? do i
> have to specify a command line flag / environment variable? what i could
> find out is that VMD needs to be configures with -P flag in order to
> allow for parallel stuff. is the fact that it correctly indicates the 8
> processors (and says, "multithreading available") already a sign for
> that it was correctly configured for multiprocessor architecture?
> thanks in advance,
> all the best,
> ben

NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email:                 Phone: 217-244-3349
  WWW:      Fax: 217-244-6078