mpich? problems on a linux cluster
c.pignedoli
c.pig... at gmail.com
Fri Dec 7 12:14:42 UTC 2007
Dear all,
I'm running cp2k on a linux cluster with the intel mpi compiler.
I'm trying to run the following (at the end of the post)
input that had no problem on a cray XT3
If I run the job on 10 nodes (each node has 2 dual core cpus and 8 GB
of memory)
I get the following error (the allocated memory is around 600 MB per
processor):
DISTRIBUTION OF THE PARTICLES (COLUMNS)
Process col Number of particles Number of matrix columns
0 123 -1
1 122 -1
2 123 -1
3 122 -1
4 122 -1
Sum 612 -1
*******************************************************************************
*** STARTING GEOMETRY
OPTIMIZATION ***
*******************************************************************************
DISTRIBUTION OF THE NEIGHBOR LISTS
Process Number of particle pairs Number of matrix elements
0 10189 1968031
1 10035 1961313
2 10289 1998317
3 9990 1963291
4 10245 1992513
5 10133 1995053
.
.
.
I cut some lines
Sum 166159 33196800
of 187578 ( 88.6 % occupation)
PSIlogger: Child with rank 3 exited on signal 11.
.
.
.
.
CP2K| Stopped by processor
number 7
CP2K| mpi_alltoallv @ mp_alltoall_c22v
CP2K| Error number
was 18
on 20 nodes the job is able to do the WFN optimization
but at the first geometry optimization step I get the following error:
Total electronic density (r-space): -3345.9999989832
0.0000010168
Total core charge density (r-space): 3345.9999999953
-0.0000000047
Total charge density (r-space):
0.0000010121
Total charge density (g-space):
0.0000010121
ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.):
-14081.720436242156211
-------- Informations at step = 0 ------------
Optimization Method = BFGS
Total Energy = -14081.7204362422
---------------------------------------------------
--------------------------
OPTIMIZATION STEP: 1
--------------------------
PSIlogger: Child with rank 76 exited on signal 11.
forrtl: error (78): process killed (SIGTERM)
Image PC Routine Line
Source
cp2k_para.x 00000000019F9A8D Unknown Unknown
Unknown
cp2k_para.x 0000000001662A61 Unknown Unknown
Unknown
cp2k_para.x 00000000016325C6 Unknown Unknown
Unknown
cp2k_para.x 00000000012D1D48 Unknown Unknown
Unknown
libmkl_lapack.so 00002B74CC848C86 Unknown Unknown
Unknown
Do you have any suggestion to solve this problem?
Thanks in advance
Carlo Pignedoli
&FORCE_EVAL
METHOD Quickstep
&DFT
BASIS_SET_FILE_NAME BASIS_SET
POTENTIAL_FILE_NAME GTH_POTENTIALS
&MGRID
CUTOFF 320
&END MGRID
&QS
EPS_DEFAULT 1.0E-12
MAP_CONSISTENT
EXTRAPOLATION PS
EXTRAPOLATION_ORDER 2
&END QS
&SCF
MAX_SCF 100
EPS_SCF 8.0E-6
MIXING 0.4
SCF_GUESS RESTART
WRITE_RESTART_EACH 100
&OUTER_SCF
OPTIMIZER DIIS
&END OUTER_SCF
&OT
MINIMIZER CG
PRECONDITIONER FULL_ALL
ENERGY_GAP 0.001
&END OT
&END SCF
&XC
&XC_FUNCTIONAL PADE
&END XC_FUNCTIONAL
&XC_GRID
XC_DERIV SPLINE2_smooth
XC_SMOOTH_RHO NN10
&END XC_GRID
&END XC
&END DFT
&SUBSYS
&CELL
A 19.98 0 0
B -9.99 17.30318757 0
C 0 0 39.673
UNIT ANGSTROM
&END CELL
&COORD
H 4.9950000000 8.6515937800 -1.0000000000
Mn -2.5048024304 8.7997909762 6.5591956999
Mn -3.6125984316 10.7073312460 6.5573654752
Mn 2.5268519596 8.8010351939 6.5480419619
Mn -1.0883339946 15.0799127199 6.5561464474
.
.
. more or less 600 atoms
.
Al 3.6838441802 15.3157889782 31.0427695918
Al 0.0034814513 8.8756390665 31.0335343629
Al -2.2900875277 12.8617932319 31.0318283490
Al 2.2900445729 12.8664182533 31.0281697464
&END COORD
&KIND H
BASIS_SET DZV-GTH-PBE
POTENTIAL GTH-PBE-q1
&END KIND
&KIND Mn
BASIS_SET DZV-GTH-PADE
POTENTIAL GTH-PBE-q15
&END KIND
&KIND Al
BASIS_SET DZVP-GTH-PADE
POTENTIAL GTH-PBE-q3
&END KIND
&PRINT
&ATOMIC_COORDINATES
FILENAME ./COORDINATE
&END ATOMIC_COORDINATES
&CELL
&END CELL
&END PRINT
&END SUBSYS
&PRINT
&FORCES
FILENAME ./FORZE_1
&END FORCES
&END PRINT
&END FORCE_EVAL
&MOTION
&CONSTRAINT
&FIXED_ATOMS
LIST 1
&END FIXED_ATOMS
&END CONSTRAINT
&GEOOPT
OPTIMIZER BFGS
MAX_ITER 400
RESTART F
&END GEOOPT
&PRINT
&FORCES
FILENAME ./FORC
&END FORCES
&STRESS
&END STRESS
&TRAJECTORY
FILENAME ./TRAJ_1
&END TRAJECTORY
&END PRINT
&END MOTION
&EXT_RESTART
RESTART_POS T
&END EXT_RESTART
&GLOBAL
FFTLIB FFTSG
RUN_TYPE GEO_OPT
PRINT_LEVEL MEDIUM
&END GLOBAL
More information about the CP2K-user
mailing list