[CP2K:1608] Re: output problems

Jun Cheng chen... at googlemail.com
Fri Dec 5 10:58:14 UTC 2008


Thanks for rapid reply.
If decreasing the size of IO buffer, will it affect the speed?
With respect to the second way, jobs normally don't stop automatically
because I am doing MD except the walltime is reached. So, cp2k may not have
time to release the buffer.

Jun

On Fri, Dec 5, 2008 at 10:44 AM, Teodoro Laino <teodor... at gmail.com>wrote:

>
> Yes, I can give at least two possibilities:
> 1) change the default of the machine on which you are running (you can
> increase/decrease the size of the IO buffer, on almost
> all machine starting from commodore64. I would suggest to get in touch
> with your system administrator for that.)
> 2) wait that the job is finished. At that point the buffer is released
> and you will have your file with all the bytes at the proper place.
>
> Teo
>
> Jun Cheng wrote:
> > Hi Teo,
> >
> > Thanks a lot for the help. It works well except the third individual
> > output file is still empty even with applying FLUSH_SHOULD_FLUSH TRUE.
> > It is probably because the third force_eva section is just a classic
> > restraint to a bond to prevent it from breaking during MD runs. So its
> > output may be too small to reach the size of the IO buffer in a
> > noticeable time. In fact, I did similar runs before and never got this
> > output in 12 hours (the walltime limit). Is there a way to overcome
> > this problem?
> >
> > Cheers,
> > Jun
> >
> >
> > On Thu, Dec 4, 2008 at 8:51 PM, Laino Teodoro <teodor... at gmail.com
> > <mailto:teodor... at gmail.com>> wrote:
> >
> >     Dear Jun,
> >
> >     1) you don't see nothing in the individual .out files because the
> >     IO is buffered.
> >     Depending on the size of the IO buffer on the machine that you are
> >     using it may take a WHILE
> >     to see something in the output file.
> >     If you want to flush you can play with this keyword:
> >
> http://cp2k.berlios.de/manual/CP2K_INPUT/GLOBAL.html#desc_FLUSH_SHOULD_FLUSH
> >
> >     2) Switching to medium you are activating some print_key, in this
> >     case DIPOLE, which for the
> >     specific force_eval you are using (MIXED) requires that DIPOLE has
> >     to be defined in all
> >     force_eval underneath. To skip this problem you can disable by
> >     hand the print_key DIPOLE:
> >
> http://cp2k.berlios.de/manual/CP2K_INPUT/FORCE_EVAL/MIXED/PRINT/DIPOLE.html
> >
> >     .....
> >         &DIPOLE OFF
> >         &END
> >     .....
> >
> >     This will solve your issue.
> >     Teo
> >
> >     On 4 Dec 2008, at 19:15, Jun Cheng wrote:
> >
> >>     Hi Teo,
> >>
> >>     I am very sorry about this.
> >>     The file (test.tgz) is uploaded in group files.
> >>     Thanks.
> >>
> >>     Jun
> >>
> >>     On Thu, Dec 4, 2008 at 6:00 PM, Teodoro Laino
> >>     <teodor... at gmail.com <mailto:teodor... at gmail.com>> wrote:
> >>
> >>
> >>         and where is your input file?
> >>         What you describe has not too much sense..anyway..
> >>         Provide your input file and all files necessary to
> >>         investigate the
> >>         problem (i.e. run it).
> >>         This is a general rule when asking for help, and I'm
> >>         surprised to see
> >>         that one has always to remind that.
> >>         Teo
> >>
> >>         Jun wrote:
> >>         > Hi all,
> >>         >
> >>         > I updated my cp2k yesterday, and did some tests on mixed multi
> >>         > force_eva calcuations. I've got some problems on the outputs:
> >>         > (i) When I used print_level low, the job got running.
> >>         Everything
> >>         > looked fine except for the empty output files corresponding
> >>         to the
> >>         > individual force_eva.(the overall output file, *ener file
> >>         and *xyz
> >>         > were fine).
> >>         > (ii) Then, I tried print_low medium. I got the individual
> >>         output
> >>         > files, but the job stopped after initialization. The error
> >>         message
> >>         > is:
> >>         > =============================================================
> >>         >  THERMOSTAT| Thermostat Info for PARTICLES
> >>         >  THERMOSTAT| Type of thermostat
> >>           Nose-
> >>         > Hoover-Chains
> >>         >  THERMOSTAT| Nose-Hoover-Chain
> >>         > length                                          3
> >>         >  THERMOSTAT| Nose-Hoover-Chain time constant
> >>         > [  fs]                      1000.00
> >>         >  THERMOSTAT| Order of Yoshida
> >>         > integrator                                       3
> >>         >  THERMOSTAT| Number of multiple time
> >>         > steps                                     2
> >>         >  THERMOSTAT| Initial Potential
> >>         > Energy                                   0.000000
> >>         >  THERMOSTAT| Initial Kinetic
> >>         > Energy                                     0.000523
> >>         >  THERMOSTAT| End of Thermostat Info for PARTICLES
> >>         >
> >>         >  ********************** begin of velocity initialization
> >>         > ***********************
> >>         >  Initial
> >>         > Temperature
> >>            330.00
> >>         > K
> >>         >  COM velocity:            0.000000000000      0.000000000000
> >>         > 0.000000000000
> >>         >  *********************** end of velocity initialization
> >>         > ************************
> >>         >
> >>         >
> >>         >  *** 17:07:15 WARNING in force_env_methods:mixed_energy_forces
> >>         > err=-300   ***
> >>         >  *** ASSERTION (cond) failed at line   1203 Error
> >>         (0.259964E-11)
> >>         > in       ***
> >>         >  *** computing numerical derivatives larger then
> >>         > (0.100000E-11) .          ***
> >>         >  *
> >>         >  *** 17:07:15 ERRORL2 in cp_result_methods:get_result_r1
> >>         > processor      0   ***
> >>         >  *** err=-300   Trying to access result ([DIPOLE]) which
> >>         was never
> >>         > stored!  ***
> >>         >  *** /home/e89/e89/jc590/work/cp2k/makefiles/../src/
> >>         > cp_result_methods.F     ***
> >>         >  *** line
> >>         > 164
> >>            ***
> >>         >  *
> >>         >
> >>         >  ===== Routine Calling Stack =====
> >>         >
> >>         >             2 qs_mol_dyn_low
> >>         >             1 CP2K
> >>         >  CP2K|  Trying to access result ([DIPOLE]) which was never
> >>         stored! /
> >>         >
> >>
> home/e89/e89/jc590/work/cp2k/makefiles/../src/cp_result_methods.F
> >>         > line    164
> >>         >  CP2K| Abnormal program termination, stopped by process
> >>         number 0
> >>         >  CP2K|  Trying to access result ([DIPOLE]) which was never
> >>         stored! /
> >>         >
> >>
> home/e89/e89/jc590/work/cp2k/makefiles/../src/cp_result_methods.F
> >>         > line    164
> >>         >  CP2K|  Trying to access result ([DIPOLE]) which was never
> >>         stored! /
> >>         >
> >>
> home/e89/e89/jc590/work/cp2k/makefiles/../src/cp_result_methods.F
> >>         > line    164
> >>         >  CP2K| Abnormal program termination, stopped by process
> >>         number 65
> >>         >  CP2K| Abnormal program termination, stopped by process
> >>         number 64
> >>         >  CP2K|  Trying to access result ([DIPOLE]) which was never
> >>         stored! /
> >>         >
> >>
> home/e89/e89/jc590/work/cp2k/makefiles/../src/cp_result_methods.F
> >>         > line    164
> >>         >  CP2K| Abnormal program termination, stopped by process
> >>         number 14
> >>         >
> >>
> ===================================================================
> >>         >
> >>         > Does anyone know what is the problem? Any reply will be
> >>         appreciated.
> >>         >
> >>         > Cheers,
> >>         > Jun
> >>         >
> >>         > P.S. I will upload my inputs if necessary.
> >>         >
> >>         > >
> >>         >
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >
> >
> >
> >
> >
> > >
>
>
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20081205/0d303010/attachment.htm>


More information about the CP2K-user mailing list