[gmx-users] Problem with continuation of files using checkpoint files

[Berk Hess]

Hi,

This means that your checkpoint file was generated with a single processor simulation,
while you are now running a two processor simulation.
This is no problem at all.
For the 4.0.4 release I changed:
"Continuation still works,"
to
"Continuation is exact,"

Note that you don't need -f and -e with tpbconv when using checkpoint.
Also you do not need tpbconv at all when you make nsteps large enough and use mdrun -maxh

Note that there is a bug in 4.0 - 4.0.3 where the initial box size is stored in the checkpoint
file with running in serial. Therefore checkpointing will give artifacts for serial NPT simulations.
NVT and parallel NPT simulations are fine.

I hope that 4.0.4, which fixes this, will be out soon.

Berk

Date: Thu, 12 Feb 2009 16:35:27 +0530
From: mon_sharma at research.iiit.ac.in
To: gmx-users at gromacs.org
Subject: [gmx-users] Problem with continuation of files using checkpoint	files

Dear All,
I am doing protein simulations using Gromacs4.0.2. For continuation or extension of runs, i am using tpbconv command. In the manual its given that for exact binary identical run, one needs to use checkpoint file with mdrun. So, for the tpbconv generated file when I am using checkpoint. I am getting warnings as below. Gromacs is  parallelly installed, but I am using on single node.  

> tpbconv_4 -s md_1.tpr -f md_1.trr -e md_1 .edr -o md_2.tpr -extend 2000.0
> mdrun_4 -s md_2.tpr -cpi md_1.cpt -deffnm md_2.tpr &

I am getting the message as:

Reading checkpoint file md_1.cpt generated: Thu Feb 12 08:42:50 2009


Gromacs binary or parallel settings not identical to previous run.
Continuation still works, but is not guaranteed to be binary identical,
 see the log file for details.

Gromacs binary or parallel settings not identical to previous run.

Continuation still works, but is not guaranteed to be binary identical.


Looking at the log file: 

Reading checkpoint file md_1.cpt
  file generated by:     mdrun_4
  file generated at:     Thu Feb 12 08:42:50 2009

  GROMACS build time:    Wed Dec 24 23:09:37 EST 2008
  GROMACS build user:    compute at localhost.localdomain
  GROMACS build machine: Linux 2.6.9-67.ELsmp x86_64
  simulation part #:     1
  step:                  500000

  time:                  2000.000000

  #PP-nodes mismatch,
    current program: 2
    checkpoint file: 1

If anyone can please suggest, what might be the problem. What is meant by PP nodes mismatch? 

Thanks in advance.

Regards
Monika




_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090212/253ce173/attachment.html>