[gmx-users] runtime statistics

[Erik Lindahl]

Hi Rui,

All the three last lines refer to nonbonded interactions with normal 
Lennard Jones and tabulated Coulomb. The first line is for interactions 
between arbitrary atoms (in most cases protein-protein), the second is 
for interactions between a protein atom and an entire water molecule 
(charge group), and the final line for interactions between two water 
molecules.

This separation doesn't affect the results; the only reason for the 
separate loops is optimization. Calling the water-water loop is much 
faster than looping over all the water atoms in the normal loop. (Two 
reasons: better cache properties, and we hide the latencies better).

Cheers,

Erik


On Monday, Feb 24, 2003, at 14:16 US/Pacific, qiao rui wrote:

> Hi, List:
> 	I am looking at the output of Gromacs (md.log) and get a little
> bit confused over the following message:
>
> Computing:        % Flop's
> ...
> Coulomb (T)       0.1
> ...
> LJ+Coulomb (T)    0.1
> LJ+Coulomb (T)(W) 1.2
> LJ+Coulomb(T)(WW) 57.8
> ....
>
> The output says that WW stands for Water-water and T for tabulated. My
> doubt is why there are LJ+Coulomb (T)(WW) and LJ+Coulomb(T)(W): is the
> latter for the interaction between Water the other molecules in the
> system?
>
> Thanks a lot for your time.
>
> Rui
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.