Error with DYNA and DOMDEC

Print Thread

#33373 02/01/14 10:56 AM

Joined: Jul 2004

Posts: 162

40.1N, 44.3E

davhak

Forum Member

davhak

Forum Member

Joined: Jul 2004

Posts: 162

40.1N, 44.3E

Dear All,

Normally using DYNA STARt with DOMDEC runs fine on C37b1 version. On one of the structures, however, it issues the following error:

Code:

 CHARMM>    DYNA cpt leap start nstep @nstep timestep @tstep -
 CHARMM>         nprint 1000 iprfrq 1000 ntrfrq 1000 -
 CHARMM>         iunrea -1 iunwri 12 iuncrd -1 iunvel -1 kunit -1 -
 CHARMM>         nsavc 1000 nsavv 0 -
 CHARMM>         PCONS pint pref 1.0 pmass @Pmass pgamma 20.0 -
 CHARMM>         TEMINC 5.0 IHTFRQ 1000 IEQFRQ 1000 -
 CHARMM>         firstt 200 finalt @temp
 Parameter: NSTEP -> "500000"
 Parameter: TSTEP -> "0.001"
 Parameter: PMASS -> "23196"
 Parameter: TEMP -> "293.0"
  IUNREA = -1         IUNWRI = 12          IUNOS = -1
  IUNCRD = -1         IUNVEL = -1          KUNIT = -1

 SELECTED IMAGES ATOMS BEING CENTERED ABOUT  0.000000  0.000000  0.000000

 NONBOND OPTION FLAGS: 
     ELEC     VDW      ATOMs    CDIElec  SHIFt    VATOm    VFSWIt  
     BYCB     NOEXtnd  EWALd   
 CUTNB  = 14.000 CTEXNB =999.000 CTONNB = 10.000 CTOFNB = 12.000
 CGONNB =  0.000 CGOFNB = 10.000
 WMIN   =  1.500 WRNMXD =  0.500 E14FAC =  1.000 EPS    =  1.000
 NBXMOD =      5
 PME EWALD OPTIONS: KAPPA  =  0.320  QCOR =  0.000  Bspline order = 6
 FFTX= 120  FFTY= 120  FFTZ= 160
                Using FFTW
                Real-to-Complex FFT
                Using Column FFT 
 There are  1516022 atom  pairs and   442585 atom  exclusions.
 There are        0 group pairs and    42721 group exclusions.
Initializing DOMDEC with NDIR =   4  4  6
Dynamic Load Balancing enabled in directions X Y Z
 DCNTRL> Constant pressure requested.
         Reference pressure tensor (XX,YY,ZZ)=        1.0       1.0       1.0 atm.

         Reference pressure tensor (XY,XZ,YZ)=        0.0       0.0       0.0 atm.

   NSTEP =   500000    NSAVC =     1000    NSAVV =        0
  ISCALE =        0   ISCVEL =        0   IASORS =        0
  IASVEL =        1   ICHECW =        1   NTRFRQ =     1000
  IHTFRQ =     1000   IEQFRQ =     1000   NPRINT =     1000
  INBFRQ =       -1   IHBFRQ =        0   IPRFRQ =     1000
  ILBFRQ =       50   IMGFRQ =       -1
  ISVFRQ =        0   NCYCLE =       50    NSNOS =       10
  FIRSTT =   200.000  TEMINC =     5.000  TSTRUC =  -999.000
  FINALT =   293.000  TWINDH =    10.000  TWINDL =   -10.000

  TIME STEP =  2.04548E-02 AKMA       1.00000E-03 PS

  RANDOM NUM. GEN. SEED(S) = 1042331256 1042331256 1042331256 1042331256

           SHAKE TOLERANCE =     0.10000E-09
 NUMBER OF DEGREES OF FREEDOM = 409157

          GAUSSIAN OPTION                  IS            1
          VELOCITIES ASSIGNED AT TEMPERATURE =    250.0000
          SEED FOR RANDOM NUMBER GENERATOR IS:
          SEEDS> 1042331256 1042331256 1042331256 1042331256

     DETAILS ABOUT CENTRE OF MASS
     POSITION          :  -6.76794775E-02  -1.55211591E-02    1.1513304    
     VELOCITY          :  -4.61475705E-04   4.29753438E-04  -1.15964161E-04
     ANGULAR MOMENTUM  :    33837.747        16402.253       -31414.070    
     KINETIC ENERGY    :   0.23839628    
mynod,j,i,is,iq=59 2548263726787167873       NaN       NaN       NaN

      ***** LEVEL -5 WARNING FROM <domdec_d2d_comm> *****
      ***** NaN coordinate
      ******************************************
      BOMLEV (  0) IS REACHED - TERMINATING. WRNLEV IS -5

                    ABNORMAL TERMINATION
                    MOST SEVERE WARNING WAS AT LEVEL -5

                    $$$$$ JOB ACCOUNTING INFORMATION $$$$$
                     ELAPSED TIME:    18.01  SECONDS 
                         CPU TIME:    16.46  SECONDS 
mynod,j,i,is,iq=56  885263726787167873       NaN       NaN       NaN

The coordinates of the system seems ok. There is no missing coordinates and also VMD opens the coordinate file without complains. Reading of the PSF and COOR files which precede the DYNA START also pass without errors.

Can someone suggest about the cause of this error?

Thanks very much for any help.

Re: Error with DYNA and DOMDEC davhak #33375 02/01/14 05:50 PM
Joined: Sep 2003 Posts: 8,660 Likes: 26 39 59 56 N, 77 06 07 W rmv Forum Member
rmv Forum Member Joined: Sep 2003 Posts: 8,660 Likes: 26 39 59 56 N, 77 06 07 W	The fact that VMD opens the COOR file w/o complaint is not all that relevant; VMD is intentionally somewhat forgiving of files that may have some defects. What happens if you try w/o DOMDEC, e.g. with 64 cores? Another thing you can try is to disable dynamic load balancing by adding DLB OFF to the command that sets up DOMDEC. Are the runs that work using the exact same script, other than the different PSF and COOR files? Rick Venable computational chemist

Moderated by lennart, rmv

Link Copied to Clipboard

CHARMM Development Project Forums User Discussion & Questions Performance and Benchmarking Error with DYNA and DOMDEC