AMBER Archive (2007)

Subject: RE: AMBER: signal SIGSEGV(11)

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Tue Apr 24 2007 - 14:43:49 CDT


Dear Pankaj,
 
It is a little difficult to remotely debug the problems you are seeing. If
you could send me a copy of the structure / prmtop / input files etc I'll
see if I can reproduce the problem. A couple of questions though. What is
your computer system - hardware / OS etc. What mpi library are you using?
 
Also have you applied all of the bugfixes for Amber 8 from the amber website
http://amber.scripps.edu/ ? You should make sure you do this and then make
clean / recompile and see if that fixes the problem.
 
You should also try compiling PMEMD in $AMBERHOME/pmemd this supports
explicit solvent simulations in parallel and is designed to use the same
input as sander and give the same output. (version 9 supports GB as well).
It should generally run faster in parallel than sander and is "different"
code - hence if you see the same problems with this then it is almost
certainly an issue with your structure (or your hardware) than a problem in
the Amber 8 sander code. If pmemd runs fine then there could a problem in
sander 8 which we could look at - although I'd like to try your system with
amber9 to see if the same problem still exists.
 
All the best
Ross
/\
\/
|\oss Walker

| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
| http://www.rosswalker.co.uk <http://www.rosswalker.co.uk/> | PGP Key
available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
 

  _____

From: owner-amber_at_scripps.edu [mailto:owner-amber_at_scripps.edu] On Behalf Of
Pankaj R. Daga
Sent: Tuesday, April 24, 2007 07:55
To: amber_at_scripps.edu
Subject: Re: AMBER: signal SIGSEGV(11)

Hi Amber community,
 
I have followed suggestion given by Chris to check if there is something
wrong about my protein file during minimization. But I found after 1300
cycles of minimization, the protein-ligand complex is fine and there is no
problem with the structure.
 
Then I tried to use only steepest descent method (ntmin = 2) and found that
the minimization with ntmin = 2 finishes without any error. But when I used
the output file for further minimization with conjugate gradient method, I
face same error once again.
 
************************************************

 MPI: On host redwood, Program /usr/local/appl/Amber8/exe/sander, Rank
 0, Process 17288 received signal SIGSEGV(11)
 
 MPI: --------stack traceback-------
 sh: line 1: idb: command not found

 MPI: -----stack traceback ends-----
 MPI: On host redwood, Program /usr/local/appl/Amber8/exe/sander, Rank 0,
Process 17288: Dumping core on signal SIGSEGV(11) into directory
/work1/r1029/HDP_APR07
 MPI: MPI_COMM_WORLD rank 0 has terminated without calling MPI_Finalize()
 MPI: aborting job
 MPI: Received signal 11
************************************************

I am still confused, if this error is related to protein structure, related
to mpi or related to sander.
 
I would appreciate if you could help me out.
 
Thanks and Regards
 
Pankaj
 
===================================================================
If your ship doesn't come in, swim out to it.......Jonathan Winters
===================================================================
Pankaj R. Daga |
Department of Medicinal Chemistry | e-mail: pdaga_at_olemiss.edu
417 Faser Hall, School of Pharmacy | fax: +1-662-915-5638
University of Mississippi | phone: +1-662-915-1853
University, MS, 38677-1848 |
===================================================================
----- Original Message -----
From: "Chris Moth" < <mailto:chris.moth_at_vanderbilt.edu>
chris.moth_at_vanderbilt.edu>
To: < <mailto:amber_at_scripps.edu> amber_at_scripps.edu>
Sent: Wednesday, April 18, 2007 3:22 PM
Subject: Re: AMBER: signal SIGSEGV(11)

> There is a chance you might find it helpful to set maxcyc to 1300 so
> that you can get final output, and take a look (visualize) the
> structure. Though the energies may look reasonable, looking at the
> structure, relative to start structure, could be informative.
> Especially look around the atoms that are listed in the output file
> often. What is going on near CG atom 4880? H1 atom 6532?
>
> If you strongly suspect the minimization algorithm is essentially
> failing, then perhaps setting input for "always use steepest descent"
> (ntmin=2) would get you to an coordinate set, from which conjugate
> gradient performs better.
>
> I'm not predicting either of the above will be successful :) - but they
> are quick things I might "try" to move ahead if I were stuck.
>
> C
>
> Pankaj R. Daga wrote:
>> Hi Dr. Case and Dr. Ross,
>>
>> Thank you very much for your inputs. I have tried to run same job with
>> single processor. But I am still facing the same error.
>>
>> The protein I am dealing with is a homology model of enzyme and it is
>> a ternary complex of enzyme-DNA-ligand. The ternary system is solvated
>> in TIP3PBOX. My aim is to run MD simulation on optimized system.
>> Hence, I started with stepwise minimization.
>>
>> First step minimization went well but when I submitted the output file
>> from first step minimization for the next step, sander minimizes the
>> molecule for few steps (1300 out of 5000). Then I face this error.
>> This error is reprodicible. I am attaching the input file and output
>> file here.
>>
>>
>> ************************************************
>>
>> MPI: On host redwood, Program /usr/local/appl/Amber8/exe/sander, Rank
>> 0, Process 17288 received signal SIGSEGV(11)
>>
>>
>> MPI: --------stack traceback-------
>> sh: line 1: idb: command not found
>>
>> MPI: -----stack traceback ends-----
>> MPI: On host redwood, Program /usr/local/appl/Amber8/exe/sander, Rank
>> 0, Process 17288: Dumping core on signal SIGSEGV(11) into directory
>> /work1/r1029/HDP_APR07
>> MPI: MPI_COMM_WORLD rank 0 has terminated without calling MPI_Finalize()
>> MPI: aborting job
>> MPI: Received signal 11
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to <mailto:amber_at_scripps.edu> amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to
<mailto:majordomo_at_scripps.edu> majordomo_at_scripps.edu
>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu