Re: [AMBER] sander number of atoms limit

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 5 Apr 2016 08:07:23 -0400

> On Apr 5, 2016, at 5:16 AM, TERREUX RAPHAEL <raphael.terreux.univ-lyon1.fr> wrote:
>
> Sorry... the exact number is the one in inpcrd files.... I have no problem with pmemd....
>
> My problem is how to minimize the system prior uses pmemd.

I think the problem here isn't that sander can't minimize or run multi-million atom systems, but rather that it has never been taught how to write 1M+ atom ASCII restart files.

There are two very simple workarounds:

1. use the CPU pmemd program to minimize instead. Pmemd.cuda minimization may also be possible if there are no very bad contacts.

2. Set ntxo=2 in the input file to sander. That will make sander write the restart file in NetCDF format which doesn't suffer the limitations that the ASCII format does.

In fact, I would suggest getting in the habit of ALWAYS using ntxo=2, since it is almost always better.

HTH,
Jason

>
>
>
> -----Message d'origine-----
> De : Gould, Ian R [mailto:i.gould.imperial.ac.uk]
> Envoyé : mardi 5 avril 2016 11:13
> À : AMBER Mailing List
> Objet : Re: [AMBER] sander number of atoms limit
>
> Dear Raphael,
>
> We routinely run systems which are greater than 1M atoms, currently our largest is 1.8M atoms, we run these on the cuda version of pmemd. You say in your e-mail that the size is exactly 1,149,859 however you inpcrd says 1,158,949 which is bigger and I think this is your problem with sander. Check that your number of atoms in the parm file you are using matches the number of atoms in the inpcrd.
>
> Cheers,
> Ian
>
> --
> Dr Ian R Gould, FRSC.
> Reader in Computational Chemical Biology Department of Chemistry Imperial College London Exhibition Road London
> SW7 2AY
> E-mail i.gould.imperial.ac.uk
> http://www3.imperial.ac.uk/people/i.gould
> Tel +44 (0)207 594 5809
>
>
>
>
>
>
>
>
>> On 05/04/2016, 09:59, "TERREUX RAPHAEL" <raphael.terreux.univ-lyon1.fr> wrote:
>>
>> Hi,
>>
>> I am trying to make a MD of a large system which exceed 1 M atoms (1,149,859 exactly). I used sander just for minimizing the system before used pmemd.
>> If pmemd and pmemd.cuda has a limit to few millions atoms, sander or its file format used seems to have 999.999 atoms limits.
>> When the minimization ended the first two lines of the crd files are changed from :
>>
>> default_name
>> 1158949
>> 80.3846739 199.1308729 115.0638892 79.5031396 199.3859663 114.6420770
>> ...
>>
>> To
>>
>> defa
>> ******
>> 80.8650585 199.3545504 114.9412061 79.8759488 199.6309180 114.8514299
>> ...
>>
>> Imin=1 seems not work on pmemd (which is logical)
>> How could minimize a system with more than 1M atoms on amber 14/ambertools 15
>> Thanks in advances
>>
>> Raphael Terreux
>>
>>
>> *************************************************************
>> Prof. Raphaël TERREUX
>> ---------------------------------------------------------
>> Tel +33 (0)4 37 65 29 43 Fax +33 (0)4 72 72 26 04
>> PRABI-LG / Laboratoire de Biologie Tissulaire et d'Ingénierie (LBTI) - UMR 5305 CNRS - Université Lyon 1.
>> Institut de Biologie et Chimie des Protéines (IBCP)
>> 7, passage du Vercors - 69367 Lyon - FRANCE
>> http://www.ibcp.fr<http://www.ibcp.fr/>
>> ----------------------------------------------------------
>> Tel: + 33(0)4 78 78 56 00 Fax: +33(0)4 78 78 56 06
>> Département pédagogique des Sciences Physico-Chimiques et Pharmacie Galénique.
>> Faculté de Pharmacie (ISPB) http://ispb.univ-lyon1.fr<http://ispb.univ-lyon1.fr/>
>> 8 avenue Rockefeller, 69373 Lyon Cedex 08, FRANCE
>> ************************************************************
>> P-HPMob
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 05 2016 - 05:30:03 PDT
Custom Search