I would also suggest you to cleverly adjust the total simulation time
to be within your overall wallclock. E.g.:
If you can only run 2.5ns, set several input files with 2.5 ns, then
you will have a restart file from the last frame of the simulation,
and with 4 inputs of 2.5ns you get the 10ns job. You can even schedule
the jobs, if possible, to start the next run when the previous is
finished.
In this way you don't really continue from a "crashed" simulation;
On 8/17/11, Bill Ross <ross.cgl.ucsf.edu> wrote:
>> I referred to the manual, and there seems to be no detail information
>> about
>> how to continue from a crashed run. For example, I want to do 10ns MD
>> simulation, but the walltime limit forced my job to stop at 2.5ns.
>>
>> I understand that I should use the restrt file from the point 2.5ns, but
>> should I rewrite the .mdin file to decrease the total simulation time to
>> 7.5ns? Or I can just use the original .mdin file?
>
> The restrt is written with some frequency, and is not written
> automatically if a job is stopped. I believe the simulation
> time is written at the beginning of the restrt file.
>
> You would want to adjust your step limit in the mdin to a value
> that could be run within the time allowed by your batch environment.
>
> Bill
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
--
Bruno Barbosa Rodrigues
PhD Student - Physics Department
Universidade Federal de Minas Gerais - UFMG
Belo Horizonte - Brazil
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Aug 18 2011 - 05:30:02 PDT
