Hi Dmitry,
This problem is known, it is caused by the 10 angstrom cutoff with a relatively small box size. If you set your first simulations cutoff to 8 angstroms it should work fine.
A bugfix addressing this problem and some other minor issues will be released on Monday. Stay tuned.
All the best
Ross
> -----Original Message-----
> From: Dmitry Mukha [mailto:dvmukha.gmail.com]
> Sent: Sunday, January 02, 2011 6:12 AM
> To: AMBER Mailing List
> Subject: [AMBER] calculations failed on GPU with ntr=1 (with bugfix 11
> applied)
>
> Amber 11 pmemd.cuda fails to get a meaningful results with the following
> input, but with the second one works fine. All available bugfixes including
> 11th were applied. The output files are attached to the mail. Surprisingly,
> all steps were completed, but with NaN bond and huge angle energy values.
> Also coordinates became infinite.
>
> Stage 4 - density equilibration
> &cntrl
> imin=0,irest=1,ntx=5,
> nstlim=25000,dt=0.002,
> igb=0, ntc=2,ntf=2,ntp=1, taup=1,
> cut=10.0, ntb=2,
> ntpr=500, ntwx=500, ntwr=25000,
> iwrap=1,
> ioutfm=1, ig=75397747,
> ntt=3, gamma_ln=2.0, temp0=300.0,
> * ntr=1*,
> &end
> Restraints for PMEMD
> 2.0
> RES 1 1
> END
> END
>
> Stage 4 - density equilibration
> &cntrl
> imin=0,irest=1,ntx=5,
> nstlim=25000,dt=0.002,
> igb=0, ntc=2,ntf=2,ntp=1, taup=1,
> cut=10.0, ntb=2,
> ntpr=500, ntwx=500, ntwr=25000,
> iwrap=1,
> ioutfm=1, ig=75397747,
> ntt=3, gamma_ln=2.0, temp0=300.0,
> * ntr=0,*
> &end
> Restraints for PMEMD
> 2.0
> RES 1 1
> END
> END
> *
> $ uname -a*
> Linux 2.6.31.5-127.fc12.x86_64 #1 SMP Sat Nov 7 21:11:14 EST 2009 x86_64
> x86_64 x86_64 GNU/Linux
> *$ cudafe --version*
> cudafe: NVIDIA (R) Cuda Language Front End
> Portions Copyright (c) 2005-2010 NVIDIA Corporation
> Portions Copyright (c) 1998-2005 Edison Design Group Inc.
> Based on Edison Design Group C/C++ Front End, version 4.1 (Nov 3 2010
> 16:25:55)
> Cuda compilation tools, release 3.2, V0.2.1221
>
> --
> Sincerely,
> Dmitry Mukha
> Institute of Bioorganic Chemistry, NAS, Minsk, Belarus
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Jan 02 2011 - 09:30:02 PST
