AMBER Archive (2009)

Subject: Re: [AMBER] Sander Stalling

From: Robert Duke (rduke_at_email.unc.edu)
Date: Mon Apr 06 2009 - 11:57:18 CDT


If you are running on a single processor and your machine does not have a
ton of memory, you may well be page faulting, which will slow performance by
probably 2 orders of magnitude or more. On redhat linux as well as most
systems there is some sort of system monitor facility that will show you
memory, cpu, disk, page fault usage. This can help you find this sort of
problem. PMEMD can run a problem of this size on a system with 2 GB for
sure, but really these larger atom systems (anything bigger than say 60,000
atoms, just off the top of my head) are best run in parallel where the
pairlist size on any one machine does not kill you.
Regards - Bob Duke
----- Original Message -----
From: "Steve Seibold" <seibold_at_chemistry.msu.edu>
To: <amber_at_ambermd.org>
Sent: Monday, April 06, 2009 12:21 PM
Subject: [AMBER] Sander Stalling

Hello AMBER

I have installed AMBER 8 on a new system. When I run sander the output
prints out to where it list the SIZE OF NONBOND LIST. At this point, I
can wait for an hour and nothing happens although sander is still
running. My worry is that my protein system with waters may be too
large. I have over 200,000 atoms with 54169900 nonbonded contacts. Would
a stalling at this point in the log be symptomatic of too large a
system? If so, I can compile AMBER to accept larger systems?

Thanks,

Steve

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber