[CP2K:1712] Re: ERRORL2 in cp_fm_types:cp_fm_create processor

Teodoro Laino teodor... at gmail.com
Wed Jan 21 16:01:42 UTC 2009


Carlo,

keep in mind that on the cray 100-150 MB (per core) are used as 
IO-buffer for lustre.
Moreover, the error you reported is clearly an error connected to an 
allocation failures.
The only way you can manage that is to try what Axel suggested:
Run in SingleNode (-SN option with yod).
In this case you should really have enough memory and don't observe the 
crash.

Teo

carlo antonio pignedoli wrote:
> Dear Axel, I would exclude that I'm running out of memory:
>
> the job (with a cutoff of 200 istead of 240) uses ~1.1 GB per core on
> my cluster on 64 cores
> (this is what I get from "top" command on the nodes).
>
> On the XRAY-XT3 I use 512 cores with 1G per core, so I expect less
> that 1G per core.
>
> How can I give you further details?
>
>
> Thanks a lot
>
> Carlo
>
> On Wed, Jan 21, 2009 at 1:08 AM, Axel <akoh... at gmail.com> wrote:
>   
>>
>> On Jan 20, 6:58 pm, "carlo antonio pignedoli" <c.pig... at gmail.com>
>> wrote:
>>     
>>> Dear all,
>>> I'm trying to do a geometry optimization on a large Au slab.
>>>
>>> On my linux cluster it works, slooooowly as obvious.
>>>
>>> If I try the same input on a CRAY-XT3 what I get is:
>>>
>>>  *** 00:46:12 ERRORL2 in cp_fm_types:cp_fm_create processor      0   ***
>>>  *** err=-300  condition FAILED at line 170                          ***
>>>       
>> my first guess would be that you are running out of memory.
>>
>> there is no paging/swapping on xt3 compute nodes and if
>> you have a large system you may just run out of memory.
>> if this is a dual-core xt3, you may try using only one core.
>>
>> cheers,
>>   axel.
>>
>>     
>>>  *
>>>
>>>  ===== Routine Calling Stack =====
>>>
>>>            10 make_preconditioner_single
>>>             9 init_scf_loop
>>>             8 scf_env_do_scf
>>>             7 qs_energies
>>>             6 qs_forces
>>>             5 cp_eval_at
>>>             4 cp_opt_gopt_step
>>>             3 geoopt_lbfgs
>>>             2 cp_geo_opt
>>>             1 CP2K
>>>
>>> I include the input for completeness.
>>>
>>> Do you have any suggestion?
>>>
>>> Thanks in advance
>>>
>>> Carlo
>>>
>>> &FORCE_EVAL
>>>   METHOD Quickstep
>>>   &DFT
>>>     BASIS_SET_FILE_NAME ./BASIS_SET
>>>     POTENTIAL_FILE_NAME ./GTH_POTENTIALS
>>>     RESTART_FILE_NAME BIGAu-RESTART.wfn
>>>     &QS
>>>       METHOD GPW
>>>       EXTRAPOLATION ASPC
>>>       EXTRAPOLATION_ORDER 3
>>>     &END QS
>>>     &MGRID
>>>       CUTOFF 240
>>>       NGRIDS 5
>>>     &END
>>>     &SCF
>>>       MAX_SCF 20
>>>       SCF_GUESS ATOMIC
>>>       EPS_SCF 1.0E-5
>>>       &OT
>>>         PRECONDITIONER  FULL_SINGLE_INVERSE
>>>         MINIMIZER  CG
>>>       &END
>>>       &OUTER_SCF
>>>         MAX_SCF 8
>>>         EPS_SCF 1.0E-5
>>>       &END
>>>       &PRINT
>>>         &RESTART
>>>           &EACH
>>>             GEO_OPT 2
>>>           &END
>>>           ADD_LAST NUMERIC
>>>           FILENAME RESTART
>>>         &END
>>>         &RESTART_HISTORY OFF
>>>         &END
>>>       &END
>>>     &END SCF
>>>     &XC
>>>       &XC_FUNCTIONAL PBE
>>>       &END XC_FUNCTIONAL
>>>     &END XC
>>>   &END DFT
>>>   &SUBSYS
>>>     &CELL
>>>       A [angstrom] 65.414448     0.000000     0.000000
>>>       B [angstrom]  0.000000    39.483733    -2.427710
>>>       C [angstrom]  0.000000     0.000000    28.0
>>>     &END CELL
>>>     &TOPOLOGY
>>>       COORD_FILE_NAME ./bigau.xyz
>>>       COORDINATE xyz
>>>     &END
>>>     &KIND Au
>>>       BASIS_SET TZ-GTH
>>>       POTENTIAL GTH-PBE-q11
>>>     &END KIND
>>>     &KIND H
>>>       BASIS_SET DZVP-GTH-PBE
>>>       POTENTIAL GTH-PBE-q1
>>>     &END KIND
>>>   &END SUBSYS
>>> &END FORCE_EVAL
>>> &GLOBAL
>>>   PRINT_LEVEL LOW
>>>   PROJECT BIGAu
>>>   RUN_TYPE GEO_OPT
>>>   WALLTIME 85000
>>> &END GLOBAL
>>> &MOTION
>>>    &CONSTRAINT
>>>     &FIXED_ATOMS
>>>       LIST  1..330
>>>     &END
>>>  &END
>>>   &GEO_OPT
>>>     MAX_ITER 5000
>>>     MAX_FORCE 0.00100
>>>     OPTIMIZER LBFGS
>>>     &BFGS
>>>     &END
>>>   &END
>>> &END
>>>
>>>       
>
> >
>   




More information about the CP2K-user mailing list