VMD-L Mailing List

From: Ondrej Marsalek (ondrej.marsalek_at_gmail.com)
Date: Tue Dec 02 2008 - 12:19:51 CST

sorry for the delay, i am trying to create a minimal script that
reproduces the problem and i get strange behavior.

axel, thanks for the explanation, i will explore this more and try to
see how much of this is an actual problem and how much is only me
being misled by my system.

so far, it seems to happen only with some data, which is not pretty.
of course, i would be the first one to blame my own code, but i don't
know how to create a memory leak from python even if wanted to :-)

ondrej

On Tue, Dec 2, 2008 at 18:23, Axel Kohlmeyer
<akohlmey_at_cmm.chem.upenn.edu> wrote:
> ondrej,
>
> please note that this is not yet a memory leak if that memory
> gets re-used for the next molecule. this is frequently a "feature"
> of the malloc(3) functionality of the respective OS and/or c++
> compiler and depends on whether the new operator gets memory
> from the stack or from the heap. on some operating systems, one
> can work around this by compiling the GNU malloc library
> (as libgmalloc.a) and link to that instead of the default malloc.
> gmalloc also uses mmap(2) to allocate (and re-use, if possible)
> larger chunks and those _can_ be given back to the process.
>
> cheers,
> axel.