VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Wed May 25 2011 - 10:18:12 CDT

Ajasja, Maria,
  Selections are a fixed size in terms of their memory footprint.
The only thing that matters in the case of a leak is how many
such selections are created, not how many atoms they contain...

In older versions of VMD that precede the debugging plugin that
Axel referred to, one can see how many atom selections exist
part way through a run like this:
  atomselect list

Cheers,
  John

On Wed, May 25, 2011 at 12:57:10PM +0200, Ajasja Ljubeti?? wrote:
> Well, I would first try to VMD 1.9 and see if it works better.
> (And if I understand correctly the new verson has a debugger plugin that
> can be very useful in exactly this situation).
> It is strange that the less waters you try to find the earlier the
> analysis crashes. If it were a memory leak one would expect just the
> opposite?
> Best regards,
> Ajasja
> On Wed, May 25, 2011 at 11:51, maria goranovic <mariagoranovic_at_gmail.com>
> wrote:
>
> Hi All
> This is vmd version 1.8.6. there is only one selection that I make using
> atomselect, and am deleting it immediately after. The number of frames
> after which i see the segmentation fault depends on the number of
> closest waters I try to find.
> If I try to find 50 closest waters, the seg fault appears after 51
> frames. Other numbers are :
> # waters # frames before segfault
> 50 51
> 100 51
> 1000 51
> 2000 51
> 3000 603
> 4000
> On Tue, May 24, 2011 at 9:08 PM, Axel Kohlmeyer <akohlmey_at_gmail.com>
> wrote:
>
> hi
>
> i totally forgot to insert a plug (pun intended) here for the vmd
> debug plugin.
>
> (sidenode: john, i just noticed that there is no working URL for the
> vmddebug plugin documentation.)
>
> On Tue, May 24, 2011 at 1:13 PM, John Stone <johns_at_ks.uiuc.edu> wrote:
> >
> > Maria,
> > This is most likely an "atom selection leak", where your script is
> > creating a large number of atom selections but not deleting them as
> > it no longer needs them. Make sure that you delete any atom
> selections
> > that are being created as each frame is loaded, before the script
> moves
> > on to process the next frame.
>
> as of VMD 1.9 there is a plugin that can help tracking how many atom
> selections
> you make and thus show you whether you have a "selection leak". it can
> even
> help with narrowing down where the leak is. this makes use of some
> advanced
> Tcl scripting features so i would love to hear, if this works for
> everybody or if not
> and where and how not.
>
> the usage of the plugin is very simple.
>
> package require vmddebug
> debug atomselect on
>
> and then run your code and with:
>
> debug atomselect stats
>
> you can get an update of how many selections have been
> allocated. if you want to track individual atomselect invocations
> and uses of the selection functions, you can do:
>
> debug atomselect verbose
>
> good luck,
> axel.
>
> >
> > Cheers,
> > John Stone
> > vmd_at_ks.uiuc.edu
> --
> Dr. Axel Kohlmeyer
> akohlmey_at_gmail.com http://goo.gl/1wk0
>
> Institute for Computational Molecular Science
> Temple University, Philadelphia PA, USA.
>
> --
> Maria G.
> Technical University of Denmark
> Copenhagen 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
http://www.ks.uiuc.edu/~johns/           Phone: 217-244-3349
http://www.ks.uiuc.edu/Research/vmd/       Fax: 217-244-6078