Help to solve OUT OF MEMORY ERROR!

rpestana94

Hi, I'm trying to do a scf calculation using Orca 5.0.1, my system is a DNA base pair (with the phosphate part) with a 1 nm gold nanoparticle (about 46 atoms), the system was previously optimized using: ! PBE D3 RI def2-SVP def2/J noautostart miniprint nopop opt
Now I'm trying to do scf using:
! PBE D3 def2-QZVP def2/J RIJCOSX noautostart miniprint nopop
%scf DirectResetFreq=1 end
%maxcore 10000


But I'm always getting this error:
[file orca_tools/qcmem.cpp, line 949]:  OUT OF MEMORY ERROR!

[file orca_tools/qcmsg.cpp, line 458]:
  .... aborting the run


I tried changing %maxcore using high values like 300000 so I don't know how to solve this, maybe if I change the basis to def2-TZVP.

Any advice to treat this problem?
Thanks in advance.

wzkchem5

It may be counterintuitive, but sometimes you have to decrease maxcore to solve an out-of-memory error.
There are two kinds of out-of-memory errors: (these do not only apply to ORCA but more or less apply to other programs as well)
(1) The memory specified by maxcore is insufficient for the calculation. In this case you should increase maxcore.
(2) The amount of memory of the machine is smaller than what would be implied by the maxcore keyword. In this case ORCA erroneously assumes that there is a lot of memory, so it chooses memory-intensive but fast algorithms over memory-efficient but slow algorithms. But when ORCA actually does the calculation, it finds that the actual memory available to ORCA is not as much as what ORCA assumed, leading to program abort. In this case you should actually decrease maxcore, so that there is the chance that ORCA chooses the memory-efficient algorithm instead and avoids the memory bottleneck.
A quick method of checking whether you are in case (1) or case (2) is: multiply maxcore by nprocs. If this is larger than 80% times the physical memory of your machine, then case (2) is very likely.

413

do not use full electron basis set for Au atom