[gmx-users] paralel scaleing

Yuguang Mu ygmu at theochem.uni-frankfurt.de
Thu Aug 7 18:21:01 CEST 2003


> PME scaling is poor, but it depends on your system size too.
>
> Most of the work is in the LJ(S) solvent loops though, so don't worry
> about the other term.
>
>
> --


I agree with you, but why the  LJ(S) parts only calculated in one cpu,
so the scaling for 4 cpu is 25% , and for 2 cpu is 50% ?




More information about the gromacs.org_gmx-users mailing list