Hi David,
Thanks for responding to my email. I tried narrowing it down. Regular MD works fine on the same cluster. The number of cpus doesn't
seem to make a difference. I ran the same simulation (SGLD) on 16 cpus then 14 cpus, I still have the same error.
Dirar
> Date: Thu, 22 May 2008 03:16:23 -0700
> From: case.scripps.edu
> To: amber.scripps.edu
> Subject: Re: AMBER: Parallel Self-guided Langevin simulation
>
> On Wed, May 21, 2008, Dirar Homouz wrote:
> >
> > I used the following input file to run a self-guided Langevin dynamics
> > simulation:
> >
> > Self-guided Langevin dynamics run
> > &cntrl
> > ntb=0,
> > imin=0, irest=0, ntx=1,
> > ntt=3, tempi=300,temp0=300.0,gamma_ln=10.0
> > ntc=1, ntf=1,
> > nstlim=50000000,
> > ntwe=100, ntwx=100, ntpr=200,
> > isgld=1, tsgavg=0.2,tempsg=1.
> > cut=999.,rgbmax=999.,
> > /
> >
> > This simulation works fine when I run it serially on my machine. However
> > when I tried to run it in parallel on a TeraGrid NCSA cluster, it crashed.
>
> Have you tried to narrow this down? For example, does it depend on running
> self-guided dynamics? Does it depend on the number of nodes you used? etc.
>
> Just seeing a segfault is not much help. But finding out which input
> parameters are correlated with crashes would be a good start.
>
> ...dac
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo.scripps.edu
_________________________________________________________________
Change the world with e-mail. Join the i’m Initiative from Microsoft.
http://im.live.com/Messenger/IM/Join/Default.aspx?source=EML_WL_ChangeWorld
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
to majordomo.scripps.edu
Received on Sun May 25 2008 - 06:07:32 PDT
