Dear all,
I am running a simulation of a small system (~200 atoms) at NPT using Amber
20 and RTX2080 cards. I used both ntt=1 and ntt=3 but the system seems to
behave strangely. I suddenly start seeing NAN, however the simulation won't
crash, but keeps running and doesn't write correct numbers in trajectory
(nc) or restart file starting from this point.
NSTEP = 5400000 TIME(PS) = 111400.000 TEMP(K) = 296.69 PRESS =
356.8
Etot = -5104.3554 EKtot = 1193.8928 EPtot =
-6298.2482
BOND = 12.1570 ANGLE = 31.3871 DIHED =
4.9381
1-4 NB = 16.6773 1-4 EEL = 309.5715 VDWAALS =
925.8803
EELEC = -7598.8595 EHBOND = 0.0000 RESTRAINT =
0.0000
EKCMT = 588.2330 VIRIAL = 430.4586 VOLUME =
20478.7190
Density =
0.9903
------------------------------------------------------------------------------
NSTEP = 5410000 TIME(PS) = 111420.000 TEMP(K) = NaN PRESS =
NaN
Etot = NaN EKtot = NaN EPtot =
NaN
BOND = 0.0000 ANGLE = 7096.0978 DIHED =
0.0000
1-4 NB = ************** 1-4 EEL = ************** VDWAALS =
NaN
EELEC = NaN EHBOND = 0.0000 RESTRAINT =
0.0000
EKCMT = -12582912.0000 VIRIAL = NaN VOLUME =
NaN
Density =
NaN
------------------------------------------------------------------------------
Once I load the trajectory, it has that many frames but can see any atoms
after exactly when this starts. I can try writing energies more frequently
but the MD trajectory and water box looks perfectly fine until this point.
Do you think it can be an issue with the GPU card? The other MD works fine
on the same card. I repeated this system with ntt=1 and with 3 but ending
up with the same issue. Any inputs?
Mish
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 03 2021 - 10:00:03 PDT
