[gmx-users] restarting a sorted shuffled trajectory

[Stéphane Téletchéa]

chris.neale at utoronto.ca a écrit :

Thanks a lot for you all, i realise i haven't posted i found the 
solution of my problem, as pointed by Yang Ye (thanks). My problem was 
that i used the  wrong .top and .gro (the sorted and shuffled one) file 
with the sorted, shuffled trr. The discrepancy came from the 
pre-processed .tpr (since one need to provide the .top and .gro linked 
into the processed .tpr tpr).

Problem solved, run done (still need a lot of work, though ;-).

However, you remark (see below) about the diminution of performance 
according to the initial sort over time is very interesting and i'll 
probably give it a try to improve scaling (on a single machine with 4 
processors, the relative performance is reported to be 78%, i would 
expect something close to 100%).

> The benefit of sort tends to deteriorate over time so I find it best to 
> do cycles of
> about 200ps. This said, shuffling and sorting my runs gives me 100% 
> efficiency to 4cpus =
> they run 4x as fast as they do on a single node. I can't come anywhere 
> near that with
> plain runs.

Thanks a lot for your time and answers,

Stéphane

-- 
Stéphane Téletchéa, PhD.                  http://www.steletch.org
Unité Mathématique Informatique et Génome http://migale.jouy.inra.fr/mig
INRA, Domaine de Vilvert                  Tél : (33) 134 652 891
78352 Jouy-en-Josas cedex, France         Fax : (33) 134 652 901