[gmx-users] Conceptual question about the computational scaling of MD

Mark Abraham mark.j.abraham at gmail.com
Sat May 11 20:53:55 CEST 2013


"for sure" is a bit of an overbid :) If the overall cost is dominated by
computation over the range of N of interest, then the scaling properties of
the computation will dominate the overall scaling.

Mark


On Sat, May 11, 2013 at 8:15 PM, francesco oteri
<francesco.oteri at gmail.com>wrote:

> Hi Andrew,
> maybe it is true on single processor simulations.
> It is false for sure when we talk about multicpu simulations because in
> this case each cpu has
> comunicate with each other and the implementations of this step makes a
> huge difference between
> the different codes.
>
>
> Francesco
>
>
> 2013/5/11 Andrew DeYoung <adeyoung at andrew.cmu.edu>
>
> > Hi,
> >
> > Please forgive me for this slightly off-topic question.  Suppose I use
> the
> > particle mesh Ewald algorithm for the computation of electrostatic
> > interactions.  It is often mentioned in the literature that the PME
> > algorithm scales as O(N logN), and that the electrostatic interactions
> are
> > the most computationally expensive part of MD.  Is it thus reasonable to
> > say
> > that MD overall scales as O(N logN)?  Or does MD overall scale in some
> > other
> > way?  I have been having difficulty finding a textbook or paper that
> states
> > or estimates the scaling of MD.
> >
> > Thank you so much for your time!
> >
> > Andrew DeYoung
> > Carnegie Mellon University
> >
> > --
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > * Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
>
>
>
> --
> Cordiali saluti, Dr.Oteri Francesco
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>



More information about the gromacs.org_gmx-users mailing list