[gmx-users] interpreting the output of gmx tune_pme

Kutzner, Carsten ckutzne at gwdg.de
Wed Jun 8 10:50:56 CEST 2016


Hi,

> On 07 Jun 2016, at 22:00, jing liang <jingliang2015 at gmail.com> wrote:
> 
> Hi,
> 
> the output of "gmx tune_pme" in perf.out file reports the following line at
> the end:
> 
> Line tpr PME ranks  Gcycles Av.     Std.dev.       ns/day        PME/f
>   DD grid
> 
> how can I interprete "Gcycles Av." and "PME/f"?
These correspond to values extracted from the md.log output
files. "Gcycles Av." is the total Giga-Cycles column from the
"real cycle and time accounting" table at the end of the log file,
averaged over the runs that tune_pme did. So it is analogous
to the total run time of the run(s).

PME/f is the computational load between the PME mesh part (on the
separate PME nodes) and the rest of the force calculation (on the
PP nodes). The more this value deviates from 1.0, the more load 
imbalance you have in that run and the less optimal it is for the 
performance.

Carsten


> 
> Thanks.
> -- 
> Gromacs Users mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.



More information about the gromacs.org_gmx-users mailing list