[Pw_forum] problems computing cholesky decomposition
YY
xrhino at gmail.com
Wed Feb 16 17:04:28 CET 2011
Dear All,
I meet a "problems computing cholesky decomposition" problem while doing the
parallel calculations. I checked the archives of this mailing list and
din't find a proper solution. So I write here for further help.
I am doing the relaxation/scf calculation. When I submitted the job on a
blue-gene machine with " -np 512 -npool 2" , it works but too slow (1000 sec
per electron step). So I tried a higher level parallel calculation with
"-np 512 -npool 2 -ntg 4 -ndiag 16", and it gave a CRASH report as:
from pdpotrf : error # 23
problems computing cholesky decomposition
I checked the memory usage, and it is not the reason.
Here is the input file and thanks for your help!
&control
calculation='scf'
restart_mode='from_scratch',
pseudo_dir = '/gpfs/small/STED/home/STEDzhyy/pseudo/pwscf'
outdir='./tmp'
prefix='pbe'
tprnfor = .true.
tstress = .false.
disk_io='none'
/
&system
ibrav= 0, celldm(1)=1.8897, nat= 249, ntyp= 4,
ecutwfc =25,ecutrho=150
occupations='smearing', smearing='gaussian', degauss=0.0072,
nspin = 2,
starting_magnetization(1)=0
starting_magnetization(2)=0
starting_magnetization(3)=0.01
starting_magnetization(4)=0.05
/
&electrons
/
CELL_PARAMETERS cubic
17.8076000000000008 -10.2812000000000001 0.0000000000000000
0.0000000000000000 20.5624000000000002 0.0000000000000000
0.0000000000000000 0.0000000000000000 25.0000000000000000
ATOMIC_SPECIES
C 12.0107 C.pbe-van_ak.UPF
H 1.00794 H.pbe-van_ak.UPF
N 14.007 N.pbe-van_ak.UPF
Cu 63.546 Cu.pbe-n-van_ak.UPF
ATOMIC_POSITIONS {crystal}
C 0.6877770687760563 0.7285346962446679 0.5652978667315952 1 1 1
C 0.6645003128091277 0.7848711188927444 0.5642487594707469 1 1 1
***********
Cu 0.0416700000000034 0.9583299999999966 0.2518400000000014 0 0 0
Cu 0.0416700000000034 0.4583299999999966 0.2518400000000014 0 0 0
K_POINTS {gamma}
Yuyang Zhang
Nanoscale Physics and Devices Laboratory
Institute of Physics, Chinese Academy of Sciences
Beijing 100190, P. R. China
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110216/3ae8e426/attachment.html>
More information about the users
mailing list