rtdb problem


Clicked A Few Times
Dear community and dear developers,

i try to run a huge DFT calculation on a BlueGene machine.
My structure consists of about 11k atoms

in order to handle the high number of atoms the following parameters were used for the compilation
parameter (nw_max_atom =15000)
parameter (nw_max_shells = 50000)
parameter (nw_max_nbf = 50000)

However, I obtain the following error

 geom_rtdb_store: write to rtdb failedgeometry -> geometry
 geom_rtdb_store: open geometies:  1
  1 geom_rtdb_store: "geometry" -> "geometry"
 geom_rtdb_store: geometries in last accessed data base:  1
 geometry
 ------------------------------------------------------------------------
 geom_input: geom_rtdb_store failed        0
 ------------------------------------------------------------------------
 An error occured in the Runtime Database


Obviously there is a problem writing to the runtime database.
Could you please give me a hint what I overlooked or tips how to solve this error ?

My kind regards

The input is the following :
scratch_dir /gpfs/scratch/userexternal/akulesza
echo

start ANIMAL1

geometry units angstrom noautoz noprint noautosym 
symmetry c1
LOAD format xyz ANIMAL.xyz 
end

basis
h,c,n,o,s,mg,cl library 6-31g*
fe,ni library cc-pvtz
end
dft
 xc becke88 perdew86
 direct
 iterations 1000
end
charge -31
task dft energy  

Clicked A Few Times
I narrowed the error by running a "Print high" job. The error now reads. Any suggestion would be greatly appreciated.
Best
Alexander

    Mapping of centers


      Group operations


  geom_rtdb_store: write to rtdb failedgeometry -> geometry
 geom_rtdb_store: open geometies:  1
  1 geom_rtdb_store: "geometry" -> "geometry"
 geom_rtdb_store: geometries in last accessed data base:  1
 geometry
 ------------------------------------------------------------------------
 geom_input: geom_rtdb_store failed        0
 ------------------------------------------------------------------------
 An error occured in the Runtime Database
 ------------------------------------------------------------------------

Forum Vet
CoolEsza
Could you upload the complete input file and xyz file to a website?

Clicked A Few Times
Gladly. Thanks for looking into this!
Here is the complete folder

input
nwchem.nw
xyz file
ANIMAL.xyz

https://drive.google.com/folderview?id=0Bx23ZKoZF1NdfnF5N01IdUhDeGtLRWRCYnF0UHhtR3JEbl9qd2...

best
Alex

Forum Vet
I can see lots of output/error files in that google drive folder ... what's the difference between them?

Clicked A Few Times
Oh I am sorry for the bad explanation:

all files are for the same job different tries (inp deck printed in the output files)

The latest run was
nwchem.nw
ANIMAL.xyz
myjob_339273.err
myjob_339273.out

ANIMAL1.sub is the submit file on FERMI (showing ressource allocation of 2048 cores)
Best
Alex

Clicked A Few Times
There is still now solution. However, i found in the error message the following statement that had slipped my mind before

rtdb_cput: abuf is too small, need=20496
0:geom_input: geom_rtdb_store failed:Received an Error in Communication
Abort(-1) on node 0 (rank 0 in comm -2080374784): application called MPI_Abort(comm=0x84000000, -1) - process 0


Please, if someone has an idea what "abuf" might be, and how I could increase it, it would be greatly appreaciated!.
Best
Alex

Clicked A Few Times
It seems that I run into several limitations of the vector sizes defined.
I have also tried with another model that has less atoms. This time i reach the DFT section at least. But it stops during with

(a bit scrambled)
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
 texas: txs_error called        0
 integer overflow,needed 443964 available 190000
 texas: txs_error called        0
 texas: txs_error called        0
 integer overflow,needed 443964 available 190000
 ------------------------------------------------------------------------
 integer overflow,needed 443964 available 190000
 integer overflow,needed 443964 available 190000
 integer overflow,needed 443964 available 190000
 integer overflow,needed 443964 available 190000
 ------------------------------------------------------------------------
 texas: txs_error called        
 integer overflow,needed 443964 available 190000
 texas: txs_error called        0
 integer overflow,needed 443964 available 190000



is there anyone that could detail me the neccessary changes before compilation in order to handle a 10ik atoms simulation in terms of the bounds for the vectors ?

Best
Alex

Clicked A Few Times
i will move this question to "compilation problems"


Forum >> NWChem's corner >> Running NWChem