From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Thu Oct 03 2013 - 12:40:05 CDT
On Thu, Oct 3, 2013 at 7:36 PM, amit banerjee
> Dear Dr. Axel Kohlmeyer,
> Thanks for your valuable suggestion. The issue here is
> not to cheat physics rather finding out some way using the configuration
> file (if at all exists) other than using GPU or multiple CPU to make the
> calculations little faster.
> And faster here do not mean accepting anything which provides wrong output
> or by any means compromising the true results .
> Reducing time cannot be done with the cost of true results.
> The question here is whether in a single quad core system is it possible to
> run two simulations simultaneously?
you cannot run two different simulations from the same input file.
of course you can run either one simulation across all four cores and
then the second or two simulations concurrently across two cores each.
most likely the first option will be faster unless you have a tiny
system, but then you would not worry about simulation time.
> Anyways, thanks for your reply.
> From: Axel Kohlmeyer <akohlmey_at_gmail.com>
> To: amit banerjee <amit_informatics_at_yahoo.com>
> Cc: namd-l <namd-l_at_ks.uiuc.edu>
> Sent: Thursday, October 3, 2013 5:39 PM
> Subject: Re: namd-l: simultaneous simulation for protein in water and SMD
> On Thu, Oct 3, 2013 at 2:33 PM, amit banerjee
> <amit_informatics_at_yahoo.com> wrote:
>> Dear Friends,
>> I have a query regarding running simulation of a
>> in water and performing SMD concurrently.
>> Although, i was able to run individual simulation, first protein in water
>> sphere/water box followed by SMD.
>> Both the simulation was tested in quad core linux and windows system
>> successfully. For 1ns simulation it took long time for each case.
>> I would like to know in a quad core single system (linux/windows) is it
>> possible concurrently running both the simulations, i.e., protein in
>> and SMD, by altering the configuration file.
>> My objective is to shorten the time period for overall simulations.
> you cannot cheat physics. well, not unless you are a rare genius. what
> use is to have results faster, if they are incorrect?
> so the only way to get your work done correctly and faster is to get
> hold of more CPUs. NAMD parallelizes very well and there are plenty of
> parallel compute resources. it only takes some effort to get hold of
>> Any helpful suggestion will be appreciated, suggestions with some .conf
>> in this regard will be additionally helpful for me.
>> looking forward for your suggestions.
>> Amit K. Banerjee
>> Indian Institute of Chemical Technology,
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0
> International Centre for Theoretical Physics, Trieste. Italy.
-- Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0 International Centre for Theoretical Physics, Trieste. Italy.
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:44 CST