On Tue, Nov 08, 2022, Prithviraj Nandigrami via AMBER wrote:
>
>I tried to run the same simulation using double precision on the GPU and
>the output seems to be fine, whereas the single precision seems to throw a
>bunch of errors (coordinate overflow etc.).
Is this near the beginning of your simulation? It may be that you have some
bad forces that will go away after a while. I'd suggest, however, doing
some MD on the CPU (as much as you can afford), and then see if the GPU
MD runs start behaving better.
>Additionally, if I set igb = 1 and cut < 999.0, pmemd does not run. The
>manual says "igb /= 0 && cut < systemsize '' is not supported in GPU. What
>is the system size being referred to here?
This is bad wording in the manual. The GPU implementation of GB doesn't
allow for cutoffs; this has nothing to do with system size. Just don't set
the cut variable when running GB, and you should be fine.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 09 2022 - 08:00:03 PST
