Error reading mo vectors depending on number of CPUs


Clicked A Few Times
Dear all,

I have been working on some dft geometry optimisations and tddft calculations with NWChem. My input file for a geometry optimisation is the following:



start geom_DP_Mes

title "DP_Mes geometry relaxation"

echo

memory total 4000 mb

geometry
C -4.31980 -3.06930 -4.25410
C -4.35710 -1.66130 -4.25630
C -3.50000 -0.93200 -3.40890
C -2.60460 -1.61060 -2.55640
C -2.56780 -3.02210 -2.55390
C -3.42580 -3.74970 -3.40410
C -1.67480 -3.69350 -1.69960
C -0.81070 -2.97820 -0.83240
C -0.84240 -1.55810 -0.84120
C -1.74540 -0.89100 -1.70670
C 0.07340 -3.67360 0.03400
C 0.93050 -2.92900 0.88660
C 0.90960 -1.50880 0.86690
C 0.02070 -0.81460 0.00530
C 1.81990 -3.59440 1.76790
C 2.68670 -2.87350 2.60920
C 2.67110 -1.46190 2.58360
C 1.78660 -0.79170 1.71930
C 3.57040 -3.55160 3.47390
C 4.43750 -2.82170 4.31060
C 4.42220 -1.41360 4.28490
C 3.53970 -0.73360 3.42290
H -4.97700 -3.62840 -4.90420
H -3.40270 -4.82970 -3.40710
H -5.04290 -1.13950 -4.90830
H -3.53400 0.14770 -3.41660
H -1.66580 -4.77340 -1.71670
H -1.79270 0.18720 -1.73100
H 1.85100 -4.67340 1.80650
H 1.79390 0.28780 1.72220
H 3.58760 -4.63150 3.49840
H 5.11430 -3.34280 4.97170
H 3.53340 0.34650 3.40890
H 5.08740 -0.85400 4.92650
C -4.08230 1.37690 4.56290
C -4.08760 2.78530 4.56400
C -3.27860 3.49410 3.65440
C -2.46320 2.79470 2.74070
C -2.45750 1.38290 2.74000
C -3.26770 0.67550 3.65250
C -1.64120 0.69130 1.82670
C -0.82550 1.38730 0.89940
C -0.83800 2.80780 0.89350
C -1.65580 3.49460 1.82590
C -0.00440 0.67140 -0.01070
C 0.79470 1.39420 -0.93480
C 0.76870 2.81440 -0.95250
C -0.04320 3.53180 -0.03430
C 1.62830 0.70530 -1.85190
C 2.42390 1.40350 -2.77820
C 2.39130 2.81470 -2.80240
C 1.56680 3.50760 -1.89720
C 3.25130 0.70330 -3.68060
C 4.04470 1.41140 -4.60430
C 4.01140 2.81910 -4.62900
C 3.18530 3.52080 -3.72950
H -4.70340 0.83340 5.26030
H -3.26890 -0.40470 3.65780
H -4.71250 3.32280 5.26230
H -3.28780 4.57430 3.66030
H -1.65340 -0.38800 1.85020
H -1.67810 4.57420 1.84690
H 1.67060 -0.37360 -1.85700
H 1.55990 4.58690 -1.93630
H 3.28160 -0.37630 -3.66780
H 4.67850 0.87340 -5.29430
H 3.16490 4.60050 -3.75300
H 4.61970 3.36180 -5.33760
C 0.10180 -5.15670 0.04800
C 1.06060 -5.86700 -0.71960
C 1.08510 -7.27640 -0.70080
C 0.15680 -7.99390 0.07520
C -0.80070 -7.29820 0.83620
C -0.82970 -5.88870 0.82890
H 1.81970 -7.81260 -1.28490
H -1.51680 -7.84940 1.42920
C -0.05860 5.01520 -0.04330
C -1.07360 5.72180 -0.73850
C 0.94220 5.75050 0.64300
C 0.92600 7.15990 0.62870
C -0.08540 7.85240 -0.06090
C -1.08430 7.13160 -0.74150
H 1.69250 7.71610 1.14890
H -1.86260 7.66330 -1.26940
C -0.09220 9.36650 -0.06540
H -0.92080 9.75960 -0.65500
H -0.18780 9.74610 0.95210
H 0.83740 9.74640 -0.48970
C -2.16710 4.97440 -1.48450
H -2.78680 5.65040 -2.07420
H -1.73640 4.23950 -2.16480
H -2.81570 4.45170 -0.78120
C 2.04920 5.03460 1.39860
H 2.66040 5.73060 1.97410
H 1.62980 4.30620 2.09260
H 2.70280 4.50980 0.70160
C 0.18870 -9.50740 0.08390
H -0.36270 -9.90950 0.93370
H -0.25770 -9.89770 -0.83030
H 1.21500 -9.86820 0.14780
C -1.87640 -5.16890 1.66340
H -2.41840 -5.85870 2.31100
H -1.41020 -4.41370 2.29700
H -2.60090 -4.67440 1.01540
C 2.07900 -5.12380 -1.56840
H 2.63910 -5.80400 -2.21090
H 1.58580 -4.39170 -2.20850
H 2.79060 -4.59900 -0.93020
end

basis spherical
  • library cc-pVDZ
end

dft
iterations 200
xc b3lyp
grid nodisk
direct
end

driver
maxiter 50
end

task dft optimize




If I run on one node then everything works fine. However, when running on more than one nodes the following error appears:

 movecs_read: failing reading from ./geom_DP_Mes.movecs
dft_gradient movecs output = ./geom_DP_Mes.movecs



------------------------------------------------------------------------
dft_gradient: could not read mo vectors 917
------------------------------------------------------------------------
------------------------------------------------------------------------
current input line :
135: task dft optimize
------------------------------------------------------------------------
------------------------------------------------------------------------
An error occured while trying to read or write to disk space
------------------------------------------------------------------------
For more information see the NWChem manual at
http://nwchemgit.github.io/index.php/NWChem_Documentation


For further details see manual section:




0:0:dft_gradient: could not read mo vectors:: 917
(rank:0 hostname:sand-2-40 pid:60808):ARMCI DASSERT fail. ../../ga-5-4/armci/src/common/armci.c:ARMCI_Error():208 cond:0
dft_gradient movecs output =
./geom_DP_Mes.movecs



Depending on the number of nodes used, the error appears at different times during the calculation. For a TDDFT calculation on the same system I got the same error in the TDDFT module, during the Davidson iterations, also in the case of using only one node. This however was solved when I used 384 CPUs for the calculation. For the geometry optimisation I used up to 512 and the error still didn't stop appearing. Please let me know if you have any suggestions on what might be causing this, thank you.

Antonis


Forum >> NWChem's corner >> Running NWChem