My Community

I would appreciate if someone could help me on Vasp.5.2.
Using the same compilation options and same input data I can not run a job using Vasp.5.2, while using Vasp.4.6 it runs well. I think is a dimensional problem, but I can't fix it. I am using:
PAW-PBE
IBRION=2
Hexagonal cell.
The problem appears when I increase the Number of Plane Waves (ENCUT). Bellow are two examples where, depend on the ENCUT value it runs well otherwise it simply stops without any error message:
Bulk system
Variables / Works / Does't works
NPLWV / 84672 / 110592
IRMAX / 1465 / 1963
IRDMAX / 3169 / 4078
NKPT / 14 / 14
IONS / 72 / 72
NGX,Y,Z / 42,42,48 / 48,48,48

Slab system
Works / Doesn't works
93312 / 128000
2056 / 2060
2292 / 3070
5 / 5
37 / 37
36,36,72 / 40,40,80
Thanks

one of the main differences between vasp.4.6 and 5.2 is the handling of the large arrays: they are loaded from stack in vasp.5.2 (which is faster), but from heap in vasp.4.6. The behaviour of your run (execution stopped without error message if the basis set sizes are increased) indicates that your stack size was too small.
please check with
ulimit -s
(some linux and unix distributions give limits to the users by default).
In any case, please set
ulimit -s unlimited
either in the rc-file of your shell (.bashrc,...) or in the script file which you use to submit your job.