From: John Stone (
Date: Thu Jan 09 2020 - 08:43:29 CST

  The best way to avoid startup overhead is to leave VMD running
and have it orchestrate larger scripts rather than the other way around.
You can have VMD launch Python and bash scripts and/or external programs
and that way there's no repeated restarting.

If turning the call sequence around is impractical for some reason, then
you CAN disable VMD attaching to the GPU(s) for CUDA/OptiX by setting
the environment variables before launching VMD:
  export VMDNOCUDA=1
  export VMDNOOPTIX=1

/bin/csh | /bin/tcsh
  setenv VMDNOCUDA 1
  setenv VMDNOOPTIX 1

Best regards,
  John Stone

On Thu, Jan 09, 2020 at 08:07:13PM +1100, Miro Astore wrote:
> Hello all,
> I have noticed that on my systems that have no access to a GPU vmd starts
> up an order of magnitude faster. When I need to script a large number of
> functions, parsing data between vmd scripts and python and bash it would
> be preferable to be able to use the CPU only type of vmd. I have tried to
> compile VMD removing flags I think are GPU related but this results in
> either the same behavior or a non working install. I've also played around
> with the environmental variables
> [1]
> I'm sure there is a better way of doing this, just wondering what I'm
> missing.
> Best, Miro
> proof: time vmd takes to start up and exit
> command: time vmd -dispdev text -e exit.tcl
> low powered cpu only laptop
> real 0m0.156s
> user 0m0.046s
> sys 0m0.048s
> workstation with GPU installed:
> real 0m2.184s
> user 0m1.993s
> sys 0m0.169s
> contents of exit.tcl: "exit"
> References
> Visible links
> 1.

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