The code doesn't yet support 12-6-4 even when ntb > 0. The problem was
that there was a way to get the cuda code engaged with 12-6-4 present in
the topology even though nothing was going to be done about it. I forget
exactly what the issue was, but I'll check back and update the master
branch.
On Thu, Jan 11, 2018 at 8:51 AM, Qinghua Liao <scorpio.liao.gmail.com>
wrote:
> Hello,
>
> I think pmemd.cuda does not support lj12-6-4 potential. I guess you have
> to use CPUs if you turn on
> this potential.
>
>
> All the best,
> Qinghua
>
> On 01/11/2018 02:49 PM, Davide Sala wrote:
> > ntb is > 0. I used the same protocol several times wothout any problem,
> > only after the last update I have started to get the error message.
> >
> > Anyway, the output is attached.
> >
> > The OS is ubuntu 16.04.
> >
> > 2018-01-11 13:52 GMT+01:00 David A Case <david.case.rutgers.edu>:
> >
> >> On Thu, Jan 11, 2018, Davide Sala wrote:
> >>> I have updated Amber17 with the last patch but now I encounter the
> >>> following bug
> >>>
> >>> :"CUDA (GPU): Implementation does not support 12-6-4 LJ potential.
> >>> The topology contains 12-6-4 coefficients and must be
> >> changed."
> >>
> >> Basically pmemd.cuda only supports 12-6-4 interactions when ntb > 0.
> >> Before
> >> update.10, the program would fail to exit properly when 12-6-4
> interactions
> >> were present and ntb=0.
> >>
> >> Notes:
> >>
> >> 1. I'm not sure this is the cause of your error: if you indeed have
> ntb>0,
> >> then we'll need more details about what other options you have selected
> in
> >> your input code.
> >>
> >> 2. cc-ing do Dave Cerutti: Dave: the error message above is not specific
> >> enough. All the other messages in update.10 indicate exactly what what
> the
> >> incompatibilty is, but this one recycles the old, non-specific error
> >> message.
> >> Can you fix this (just in git master, I think)?
> >>
> >> ....dac
> >>
> >>
> >> _______________________________________________
> >> 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 11 2018 - 11:00:02 PST
