Re: [AMBER] AMBER20 GPU error

From: Thomas Cheatham via AMBER <amber.ambermd.org>
Date: Thu, 3 Nov 2022 16:17:59 -0600 (MDT)

> Thank you for the helpful suggestions. I am doing tests by reducing the force constant values.
>
> I also encountered another error for some of the MD frames:
>
> Error: '*' encountered (atom 6 X). This indicates coordinate overflow.
> Error: '*' encountered (atom 6 Y). This indicates coordinate overflow.
> Error: '*' encountered (atom 6 Z). This indicates coordinate overflow.
> Error: '*' encountered (atom 7 X). This indicates coordinate overflow.
> .........

(1) try on CPU first

(2) coordinate overflow means you are blowing the ascii format, should use
netcdf for restart files

ioutfm = 1,

(3) 100.0 restraint_wt may be too high; try something like 1 or 5

(4) If it is blowing up quickly, dump mdout and trajectory every step and
"watch" what is happening to atom's 6 and 7.

>     igb = 1, cut = 999.0,
>     iwrap = 1

You cannot wrap a GB simulation so iwrap = 0,

--tec3
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 03 2022 - 15:30:03 PDT
Custom Search