insufficient virtual memory in the preparatory DFT of G0W0 calculation

[njuxyh]

Hi :
my system only contains 4 C atoms.
and i want to do a G0W0 calculation. according to the mamual of VASP at
http://cms.mpi.univie.ac.at/vasp/vasp/R ... tions.html
i split the DFT calculation into two steps. But at the second DFT calculation (HERE the XC FUNCTIONAL is VDW-FUNCTIONAL).the code reports
insufficient virtual memory
9.13719988157784 10.2211985519339 11.4050421708962
12.6979272569059 14.1098974013170 15.6519212905873
17.3359779149375 19.1751496254094 21.1837237622236
23.3773036439213 25.7729297794946 28.3892122451236
31.2464752538692 34.3669150413914 37.7747722942059
41.4965204599610 45.5610714025986 50.0000000000000
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
Total vdW correction in eV: 4.18603611853080
forrtl: severe (41): insufficient virtual memory
Image PC Routine Line Source
vasp 0000000000D3CCCD Unknown Unknown Unknown
vasp 0000000000D3B7D5 Unknown Unknown Unknown

MY INCAR of the second DFT
PREC=A
ENCUT=500
EDIFF=1E-5
ISMEAR=0
SIGMA=0.05
#NEDOS=1001
NBANDS=24
#EDIFFG=-0.005
#ISIF=2
#IBRION=2
#NSW=150
#NPAR=12
#ADDGRID=.T.
LWAVE=.T.
LCHARG=.T.
#NELMIN=8
ALGO = Exact
NELM = 1
LOPTICS = .TRUE.

AND THE K is 21*21*1
My conpute node has 98994016k MEM, so what is wrong with it ?
<span class='smallblacktext'>[ Edited ]</span>

[admin]

I would suggest to decrease the k-mesh