[gmx-users] Checkpointing

[Mark Abraham]

On 03/02/10 15:44, Jack Shultz wrote:
> I will try it without the checkpointing flags.

That's not quite what I said. I suggested not using *different* 
filenames for -cpo and -cpi. What you want is the output file from a 
former run to transparently become the input file for the subsequent 
one, and I expect GROMACS will handle this for you if you let it. Try 
some things and see.

Mark

> If that fails, maybe
> I'll introduce some python into this integration. Check if the
> checkpoint already exists.
>
> On Tue, Feb 2, 2010 at 6:18 PM, Mark Abraham<mark.abraham at anu.edu.au>  wrote:
>>
>>
>> ----- Original Message -----
>> From: Jack Shultz<js at drugdiscoveryathome.com>
>> Date: Wednesday, February 3, 2010 2:36
>> Subject: [gmx-users] Checkpointing
>> To: Discussion list for GROMACS users<gmx-users at gromacs.org>, Andrey Voronkov<av at drugdiscoveryathome.com>
>>
>>> We have mdrun integrated into our distributed computing project. When
>>> your users suspend or close the manger it checkpoints, so when they
>>> open again it continues mdrun where it left off. However, when users
>>> reboot, it starts from the beginning. We are using this command line
>>> to execute the work.
>>>
>>> mdrun.exe (-v -x -c -o md.pdb -e -cpo md.next -cpi md.cpt -
>>> deffnm md)
>>
>> If you're using this input line always, then you're getting what you're asking for - an mdrun that begins from the state in md.cpt. Since you're never updating that, it doesn't change.
>>
>> Try not stipulating different -cpo and -cpi and see what the native coping mechanism is. Otherwise, you'll have to write a bunch of scripting logic to decide which .cpt to use each restart.
>>
>> Mark
>>
>>> I have a seperate checkpoint for output so after this simulation we
>>> can extend the workunit. Should we try using this append option?
>>>
>>> Checkpoints containing the complete state of the system are
>>> written at
>>> regular intervals (option -cpt) to the file -cpo, unless option -cpt
>>> is set to -1. A simulation can be continued by reading the full state
>>> from file with option -cpi. This option is intelligent in the
>>> way that
>>> if no checkpoint file is found, Gromacs just assumes a normal
>>> run and
>>> starts from the first step of the tpr file.
>>>
>>> With checkpointing you can also use the option -append to just
>>> continue writing to the previous output files. This is not
>>> enabled by
>>> default since it is potentially dangerous if you move files, but if
>>> you just leave all your files in place and restart mdrun with exactly
>>> the same command (with options -cpi and -append) the result will be
>>> the same as from a single run. The contents will be binary identical
>>> (unless you use dynamic load balancing), but for technical reasons
>>> there might be some extra energy frames when using checkpointing
>>> (necessary for restarts without appending).
>>>
>>>
>>>
>>>
>>> --
>>> Jack
>>>
>>> http://drugdiscoveryathome.com
>>> http://hydrogenathome.org
>>> --
>>> gmx-users mailing list    gmx-users at gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> Please search the archive at http://www.gromacs.org/search
>>> before posting!
>>> Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-users-request at gromacs.org.
>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
>
>
>