I'd copy/paste that abnormal termination with cuda, plus how you're
equilibrating (mdin files).
Possibly if you equilibrate on CPU for a little, it'll run on GPU.
Bill
On 12/12/20 10:46 AM, Amit Sharma (Asstt. Prof., MCARS) wrote:
> Hi,
>
> In an NVT simulation I want to keep the time step as 1fs and SHAKE on all
> bonds including hydrogen.
>
> Please clarify if this means keeping ntc = 3 & ntf = 3? Or, if ntc and ntf
> should be specified with some other values?
>
> When I run NVT using ntc=ntf=3, run using pmemd.cuda gets terminated
> abnormally. However, it works fine with sander, but it is too slow. Any
> suggestion if we can use cuda when ntc=ntf=3?
>
> Thanks,
>
> Amit
>
--
Phobrain.com
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Dec 12 2020 - 11:00:02 PST