Dear Amber community,
the problem still persists with the latest update (13). I've done some
more experimenting. I'm running with Ubuntu 14.04, CUDA 6.5 and Nvidia
drivers 346.96.
The test.cuda runs fine.
The pmemd.cuda simulation fails reliably.
Identical simulation on pmemd.MPI completes without any problems.
I'm simulating DNA and the problem occurs when I combine dihedral
restraints for the backbone in the following format
&rst iat = 29,30,33,34, r1 = -121.0, r2 = -120.0, r3 = 120.0, r4 =
121.0, rk2 = 200.0, rk3 = 200.0, &end
and base planarity restraints in a different format
&rst iat=29,29,1,1, atnam(1)='N3', atnam(2)='N1', atnam(3)='N3',
atnam(4)='C6', r1=350.0, r2=355.0, r3=365.0, r4=370.0, rk2 = 25.0, rk3=
25.0, iresid=1, &end
Only one or the other work fine, but both sets of restraints combined
cause problems. I will try to do some more experimenting in the coming
days.
Best, Peter
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Nov 13 2015 - 06:00:05 PST
