AW: Two GPU-based workstation

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Tue Oct 29 2013 - 05:00:15 CDT

And it's always worth to try "twoawayx yes" for small systems.

 

Norman Geist.

 

Von: Ajasja Ljubeti [mailto:ajasja.ljubetic_at_gmail.com]
Gesendet: Dienstag, 29. Oktober 2013 10:46
An: Norman Geist
Cc: Shubhra Ghosh Dastidar; Namd Mailing List
Betreff: Re: namd-l: Two GPU-based workstation

 

5. You can of course run one simulation using all the cores and both
GPUs simultanously, what is the best scenario here IMHO.

 

That depends strongly on the system size (and hardware configuration). For
small systems (~50k atoms) it might be advantageous to run parallel
simulations on each GPU card (while not oversubscribing the CPU). I usually
just run some benchmarks before starting long simulations...

I completely agree with points 1-4 however:)

 

Best regards,

Ajasja

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:49 CST