AMBER Archive (2009)
Subject: Re: [AMBER] Verlet update time in explicit water Replica Exchange MD
From: Da-Wei Li (lidawei_at_gmail.com)
Date: Wed Nov 25 2009 - 13:43:58 CST
You are absolutely right. The reason is that I do not designate the
Velocity output file, therefore all replicas try to write to same file
and cause jam. This is not obvious for implicit simulaitons.
( BTW, you cannot use NTWPRT in REMD, otherwise sander will crash when
try to exchange. )
Have a nice Thanksgiving.
On Wed, Nov 25, 2009 at 12:38 PM, Robert Duke <rduke_at_email.unc.edu> wrote:
> I am not a replica exchange expert, but I am presuming the master nodes
> might be writing to disk? Could it be that if you have the former master
> arrangement, you are writing to disk for some of the copies over an nfs
> share? This can be really bad, especially if the nfs share is either
> heavily loaded, or uses the same interconnect as mpi. I am just guessing
> here, as this is sander, and replica exchange, neither of which I do all
> that much with...
> Regards - Bob Duke
> ----- Original Message ----- From: "Da-Wei Li" <lidawei_at_gmail.com>
> To: "AMBER Mailing List" <amber_at_ambermd.org>
> Sent: Wednesday, November 25, 2009 12:28 PM
> Subject: Re: [AMBER] Verlet update time in explicit water Replica Exchange
>> Just another update. If I re-arrange the MPI machine file to let the
>> master processors of each group to be on the same node, everything
>> will be fine. I have no idea why this happens.
>> have a nice day.
>> AMBER mailing list
> AMBER mailing list
AMBER mailing list