The intention of restart files is that they will be used in precisely this
manner. MD calculations can take weeks for a long run--there is no
guarantee that the computer will stay up for that long, so the checkpoint
(restart) files are there to save progress. The only issue you need to
worry about is that you are not restarting the runs with the same random
number seed (igseed) every time--but even if you were, if you are doing
50ns at a time then the effects should be negligible.
Dave
On Mon, Aug 19, 2019 at 8:23 AM Daniel Roe <daniel.r.roe.gmail.com> wrote:
> Hi,
>
> On Mon, Aug 19, 2019 at 6:57 AM Midhun K Madhu <midhunk16.iiserb.ac.in>
> wrote:
> > My question is, is it fine to consider the restart file(.rst) generated
> at
> > the end of the first step to start the next step using pmemd.cuda
> program?
> > Will it be a continuous run? As there is another option of
> pmemd.cuda.DPFP
> > which will give more accurate restart files, I am a bit concerned about
> the
> > restart files generated using pmemd.cuda.
>
> The (default NetCDF) restarts from pmemd.cuda are written with full
> double precision, just like the restarts from CPU pmemd; there really
> is no "more accurate" than that. Is there a specific reason you are
> concerned about restart files from pmemd.cuda? Have you seen strange
> behavior etc?
>
> >
> > Also, what is the difference between pmemd.cuda and pmemd.cuda.SPFP?
>
> Typically the former is a soft link to the latter. The SPFP refers to
> the single-precision/fixed-precision model - for more details see
> https://www.sciencedirect.com/science/article/pii/S0010465512003098.
>
> -Dan
>
> >
> > regards,
> >
> > --
> >
> > *MIDHUN K MADHU*
> > Ph.D Student
> > Dept. of Biological Sciences
> > IISER Bhopal
> > _______________________________________________
> > 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 Mon Aug 19 2019 - 10:00:01 PDT
