AMBER Archive (2008)

Subject: Re: AMBER: minimization stops or does not write anything

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu Aug 14 2008 - 07:39:01 CDT

Looks to me like you don't have a periodic boundary system in the inpcrd
files, as there is no box information. Did you perhaps forget to solvate in
xleap when you built the prmtop and inpcrd files? By default, you are
running pme, which requires periodic boundary conditions. The tipoff is the
two lines in the mdout below:
> | peek_ewald_inpcrd: Box info not found in inpcrd
> |Largest sphere to fit in unit cell has radius = -0.021
Now, you say you have more than 700,000 atoms - sort of makes me wonder if
you really don't have the solvent. At any rate, somehow box info is not
getting read in the inpcrd, and the only explanation I have based on the
input here is that it does not exist.
As to system size, this is indeed a large system, but the limit for amber
10, 9, and probably 8 is 999,999 atoms. Now, I would run a system like this
with at least 1 cpu per 100,000 atoms, just to be sure I did not hit memory
problems (there are nasty things that can happen with sander in mpi, where
multiple cpu's share the same memory, whereby two cpu's both try to allocate
memory for mpi but there is not enough memory to complete both requests, and
the system hangs, or deadlocks, on what is called a "race" condition). This
problem is much worse for sander than pmemd; pmemd requires less memory, and
does mpi differently. But I think before the issue of system size comes up,
you have to work out why you don't have a box data in the inpcrd.
Regards - Bob Duke

----- Original Message -----
From: <rebeca_at_mmb.pcb.ub.es>
To: <amber_at_scripps.edu>
Sent: Thursday, August 14, 2008 5:01 AM
Subject: AMBER: minimization stops or does not write anything

> Hello,
> I am trying to minimize a membrane+protein system (solvated and with ions)
> that
> has more than 700,000 atoms.
> In a local cluster, the calculation begins, but then it is not kill, but
> stops
> writting in this point:
>
> File Assignments:
> | MDIN: min_ntr_50_all.in
> | MDOUT: min_ntr_50_all.o
> |INPCRD: equilibrado_amber.crd
> | PARM: equilibrado_amber.top
> |RESTRT: min_ntr_50_all.r
> | REFC: equilibrado_amber.crd
> | MDVEL: mdvel
> | MDEN: mden
> | MDCRD: mdcrd
> |MDINFO: mdinfo
> |INPDIP: inpdip
> |RSTDIP: rstdip
>
>
> Here is the input file:
>
> #minimisation of waterbox with restraints on all protein atoms
> &cntrl
> imin=1, maxcyc=5000, ncyc=5000, drms=0.01, ntb=1, ntr=1,
> /
> Group input for restrained atoms
> 50.0
> RES 1 2565
> END
> END
>
>
> --------------------------------------------------------------------------------
> 1. RESOURCE USE:
> --------------------------------------------------------------------------------
>
> | Flags:
> getting new box info from bottom of inpcrd
> | INFO: Old style inpcrd file read
>
> | peek_ewald_inpcrd: Box info not found in inpcrd
> |Largest sphere to fit in unit cell has radius = -0.021
>
>
> When I tried it in another cluster (Marenostrum), it does not even begin
> to
> write anything. The calculation is not killed, it creates an empty output
> file,
> but does not write anything on it.
>
> Does anybody know what could be happening? Is there any atom limit
> calculation
> to calculate in Amber?
>
> Thank you very much for your help.
>
> PD: I am using Amber9
>
> Rebeca García Fandiño
> Parc Cientific de Barcelona
>
> The calculation does not write anything
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo_at_scripps.edu
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu