The ABINIT Users Mailing List ( CLOSED )

[<martin-callsen@gmx.de>]

So here is the complete input-file for the case where the problem occured:

==================================================================

chkprim 0
chkexit 1
occopt 3
tsmear  0.005

paral_kgb 1
nloalg 4                        # comment see below
wfoptalg 14
fft_opt_lob 2   
fftalg 401
npband 2        
nbdblock 2
npfft 1
npkpt 1

ionmov  1
ntime   1000
dtion 50
vis   100
tolmxf  5.0d-4
natfix  24
iatfix  13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 
36

acell    16.602937932938254  19.171421369841099  27.11248411

ntypat 3
znucl 6 1 47

nband  184

natom 36
typat 6*1 6*2 24*3

xcart
              0.0000000000E+00  2.6380576815E+00  0.0000000000E+00
              2.2846788947E+00  1.3190288408E+00  0.0000000000E+00
              2.2846788947E+00 -1.3190288408E+00  0.0000000000E+00
              0.0000000000E+00 -2.6380576815E+00  0.0000000000E+00
             -2.2846788947E+00 -1.3190288408E+00  0.0000000000E+00
             -2.2846788947E+00  1.3190288408E+00  0.0000000000E+00
              0.0000000000E+00  4.6846310834E+00  0.0000000000E+00
              4.0572420073E+00  2.3432604048E+00  0.0000000000E+00
              4.0572420073E+00 -2.3432604048E+00  0.0000000000E+00
              0.0000000000E+00 -4.6846310834E+00  0.0000000000E+00
             -4.0572420073E+00 -2.3432604048E+00  0.0000000000E+00
             -4.0572420073E+00  2.3432604048E+00  0.0000000000E+00
   0.0000000000000000        0.0000000000000000        9.0360000000000000
   2.7671563221563753        4.7928553424602738        9.0360000000000000
   0.0000000000000000        9.5857106849205493        9.0360000000000000
   2.7671563221563753        14.378566027380824        9.0360000000000000
   5.5343126443127506        0.0000000000000000        9.0360000000000000
   8.3014689664691268        4.7928553424602738        9.0360000000000000
   5.5343126443127506        9.5857106849205493        9.0360000000000000
   8.3014689664691268        14.378566027380824        9.0360000000000000
   11.068625288625501        0.0000000000000000        9.0360000000000000
   13.835781610781876        4.7928553424602738        9.0360000000000000
   11.068625288625501        9.5857106849205493        9.0360000000000000
   13.835781610781876        14.378566027380824        9.0360000000000000
   2.7671563221563753        1.5976184474867581        4.5187473518664438
   5.5343126443127506        6.3904737899470323        4.5187473518664438
   2.7671563221563753        11.183329132407307        4.5187473518664438
   5.5343126443127506        15.976184474867582        4.5187473518664438
   8.3014689664691268        1.5976184474867581        4.5187473518664438
   11.068625288625501        6.3904737899470323        4.5187473518664438
   8.3014689664691268        11.183329132407307        4.5187473518664438
   11.068625288625501        15.976184474867582        4.5187473518664438
   13.835781610781876        1.5976184474867581        4.5187473518664438
   16.602937932938250        6.3904737899470323        4.5187473518664438
   13.835781610781876        11.183329132407307        4.5187473518664438
   16.602937932938250        15.976184474867582        4.5187473518664438

ecut  8.0
pawecutdg 16

ixc 11

kptopt 0
nkpt 1

nsym 1

nstep 100
toldfe 5.0d-6

diemac 2.0
dielng 0.5
diemix 0.3 

==========================================================================

The following comment belongs to nloalg = 4 in the input and can be skipped, 
if
this is not of interest.
As far as i understood the help-file nloalg should be 14, if i want to use
band/FFT parallelisation in the paw case. But then i get the error message:

chkinp : ERROR -
  The value of nloalg is found to be   1
  (k+G) vectors have to be precomputed in the case of band-FFT 
parallelization.
  Action : put nloalg < 10 in your input file (try nloalg 4)

In my opinion this is because in the chkinp.F90  nloalg := nloalg(5)  is
checked to be /= 0, so nloalg >= 10 
is treated as noalg = 1. But if i comment this line in the chkinp.F90 and use
nloalg 14 the memory problem does not change.



machine and compiler information:
i have downloaded the abinit-5.8.3 source-package on 30 .06. 2009 and built it
with ./configure --enable-mpi --with-mpi-prefix=/opt/mpich2 --disable-etsf-io
with gfortran 4.1.2 and the corresponding 
mpich2-1.0.8p1 on an  dual-amd-opteron machine with 4GB of memory.
I have read in the known_problems that 4.1.2 might cause problems, so i tried
the newer 4.3.1, but that did not change the memory problem. Only the
build-problem with the etsf_io-plugin dissappeared.


I hope the  error is somewhere in my input-file and not within the code or
related to the architecture. 

Martin Callsen