The ABINIT Users Mailing List ( CLOSED )

[<martin-callsen@gmx.de>]

Hello,

when i do parallel calculations using abinip-5.8.3 and set the appropriate
inputvariables as described in the help-files ( i.e. paral_kgb = 1, nloalg =
14, wfoptalg = 14, fftalg = 401 and fft_opt_lob =2) the memory needed for the
job grows slow but continuous until the job aborts. For example a job needing
about 200 MB in the sequential case gets out of memory on a 4GB machine after
13h of calculation time.

I have compiled abinit on an opteron system using gfortran and mpich2. 

Did anyone else have a similar problem before? My search on the forum would
answer this question with no.

I can send the hole input-file on request (maybe i am doing something 
generally
wrong), but i did not want this post to get too long.

Thank you for your help,
Martin Callsen