VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Wed Jan 20 2010 - 12:46:35 CST

Hi,
  It is best to ask about NAMD related issues on the NAMD-L mailing list.
My thought is that if you're not getting any output from NAMD
(your comment about an empty log file), then it's likely that something
is failing very early in the startup process when the NAMD processes
are launched on the other nodes. Have you run any of the Charm++
"megatest" self-tests on this system? I would suggest running the
Charm++ tests on your machine and if you have problems with those, you'll
want to report it to the Charm++ developers here:
  ppl_at_cs.uiuc.edu

I'm CCing the Charm++ developers on this note in case they have
additional suggestions.

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Wed, Jan 20, 2010 at 01:49:43PM +0100, Carsten Olbrich wrote:
> Hi,
>
> we observe quite the same problem with NAMD_2.7b2_Linux-x86_64 (no ibverbs).
> When we run our simulation with about 100000 atoms on 40 cpus
> everything runs fine. Using more than 40 cpus we get an empty logfile
> only, although everything runs fine using the same setup with
> NAMD_2.6_Linux-x86_64.
>
> Do you have an idea or comments about this?
>
> Best,
> Carsten
>
> On Wed, Jan 20, 2010 at 10:11 AM, Jesper Soerensen <lists_at_jsx.dk> wrote:
> > Hello,
> >
> > I have downloaded the binary "NAMD_2.7b2_Linux-x86_64-ibverbs.tar.gz".
> >
> > I've installed it and it works fine for part of our cluster.
> >
> > However, we have recently added new machines, where suddenly nothing is
> > written to the logfile and no other files are generated. We are using a
> > newer version of OFED on these machines (version 1.5) and that is really
> > the only change.
> >
> > The jobs execute correctly on one node, but once i go for two nodes the
> > jobs starts and I can see several namd processes running on the two
> > nodes, but nothing is written to the filesystem not the log-file or
> > anything else. The logfile is touched initially, but nothing is written
> > to it.
> >
> > Can this be because of the OFED version?
> >
> > Kind regards,
> > Jesper Soerensen
> >
> > 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078