[CP2K-user] Error in MPI when running a QMMM simulation
Tianshu Jiang
jts2t... at gmail.com
Tue Dec 3 02:32:01 UTC 2019
Hi everyone in cp2k community,
I'm using cp2k4.1 to run a QMMM calculation, but the strange thing is that
different result emerged under 1 message passing process and
72 message passing processes situation. I run same example using 1 core and
72 cores simultaneously. But after 2000 steps, the program
crashed in 72 cores' simulation. And the error report says
#2 0x2AD0BBB5A24F
#3 0xAC14E1 in __input_cp2k_restarts_util_MOD_section_velocity_val_set
#0 0x2AF10E6856F7
#2 0x2B0AD401624F
#3 0xAC14E1 in __input_cp2k_restarts_util_MOD_section_velocity_val_set
#3 0xAC14E1 in __input_cp2k_restarts_util_MOD_section_velocity_val_set
#2 0x2AF10F55924F
#4 0x9EE04F in __input_restart_force_eval_MOD_update_subsys
#4 0x9EE04F in __input_restart_force_eval_MOD_update_subsys
#4 0x9EE04F in __input_restart_force_eval_MOD_update_subsys
#0 0x2B06FD3176F7
#2 0x2B06FE1EB24F
#2 0x2AE46DD2C24F
#5 0x9EF0B3 in __input_restart_force_eval_MOD_update_force_eval
#0 0x2AC0F9FB96F7
#1 0x2AC0F9FB9D3E
#5 0x9EF0B3 in __input_restart_force_eval_MOD_update_force_eval
Setting about restart part reads
&EXT_RESTART
RESTART_FILE_NAME ../equi_qmmm/UREA-ZW-1_20000.restart
RESTART_DEFAULT F
RESTART_POS
RESTART_VEL
&END
In this simulation, I choose GPW as the electronic structure method. Part
of relative setting reads
&DFT
CHARGE 0
BASIS_SET_FILE_NAME BASIS_SET
POTENTIAL_FILE_NAME GTH_POTENTIALS
&QS
METHOD GPW
EPS_DEFAULT 1.0E-10
&END QS
&XC
&XC_FUNCTIONAL PADE
&END XC_FUNCTIONAL
&END XC
&SCF
MAX_SCF 30
EPS_SCF 1.0E-6
SCF_GUESS ATOMIC
&OT
MINIMIZER DIIS
PRECONDITIONER FULL_SINGLE_INVERSE
&END
&OUTER_SCF
EPS_SCF 1.0E-6
MAX_SCF 10
&END
Prior to this simulation, I also run same example using PM6 as QM method.
But there is no such problem even under MPI running.
So where is the problem ? Could someone give any suggestion ?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191202/4c3097f7/attachment.htm>
More information about the CP2K-user
mailing list