[AMBER] NMR restraints on GPU (asterisks in output)

From: Miroslav Krepl <krepl.seznam.cz>
Date: Fri, 17 Jun 2016 11:21:15 +0200

Dear AMBER,

I have been testing the new GPU based NMR restraints simulations (Kudos
for that, btw!) and I encountered a rather strange error in my output
files.

Specifically, at random points, I get a strange output printed like this
(the asterisks being the strange part), e.g.: 

----
 NSTEP =    15000   TIME(PS) =   16160.000  TEMP(K) =   301.41  PRESS =
0.0
 Etot   = **************  EKtot   =     10693.0312  EPtot      =
**************
 BOND   =       256.3677  ANGLE   =       440.9956  DIHED      =
662.9386
 1-4 NB =       245.4594  1-4 EEL =     -2889.7832  VDWAALS    =
10976.7385
 EELEC  =    -81844.9522  EHBOND  =         0.0000  RESTRAINT  =
**************
 EAMBER (non-restraint)  =    -72152.2357
 EKCMT  =         0.0000  VIRIAL  =         0.0000  VOLUME     =
175110.6222
                                                    Density    =
1.0483
------------------------------------------------------------------------------
 NMR restraints: Bond =*********   Angle =     0.000   Torsion =
0.000
----
On CPU, this usually means that the restraint energy somehow rose to
(near) infinity and the simulation is blowing up. But in this case, the
thing is that a next output step, it is usually normal again (with no
asterisks). Also, the resulting trajectory looks normal to me without
any missing frames or anything. There are also no error messages to
standard input or output.
It seems that the asterisks in the output are really not connected to
anything and so I am not sure whether to be concerned.
This is not happening with the pmemd.MPI CPU based version. It is also
not happening with pmemd.cuda when not using NMR restraints. 
Could this be a bug in the output printing routine?
My gpu is a GTX Titan Black.
----
Anyway, this is my AMBER input file:
MD production
 &cntrl
    imin = 0, nmropt = 1, ntx = 5, irest = 1,
    igb = 0, ntb = 2, ntpr = 1000, ntwx = 2500,
    cut = 9.0, ioutfm=1, ntxo=2,
    tempi = 300.0, temp0 = 300.0,
    ntt = 1, tautp = 1.0, ig=-1,
    ntp = 1, taup = 1.0, barostat=2,
    nstlim = 250000, dt = 0.004,
    ntc = 2, ntf = 2, tol = 0.00000001,
    ntwr = 10000, nscm = 2500,
    iwrap = 0,
 &end
 &wt type='REST', istep1=0, istep2=250000, value1=RK0, value2=RK1, /
 &wt type='END' /
DISANG=RST.DISANG
----
and this is my setting in the DISANG file:
rk2=32.0, rk3=32.0, ir6=1, ialtd=0,
----
Thank you for your advice!
With best wishes,
Miroslav
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jun 17 2016 - 02:30:02 PDT
Custom Search