Author Topic: $rpacor $maxcor $ricore questions  (Read 711 times)

inozerox

  • Jr. Member
  • **
  • Posts: 17
  • Karma: +0/-0
$rpacor $maxcor $ricore questions
« on: June 11, 2016, 12:06:14 am »
Hello everyone.

I have some questions related to $rpacor $maxcor and $ricore.

Do all of them use a RAM memory? Hard disk?

Which combinations of them is the most convenient in order to perform the following calculations?

Optimizing the first singlet excited state, performing a TD-DFT calculation, Geometry Optimization using SCS-MP2 level (ricc2) and Geometry Optimization using a Hybrid Functional (DFT).

I apologize for asking a lot. But, I have been using Turbomole since 3 years ago and I just realize that maybe I have been making a mistakes respect to memory assignment.

Thanks in advance

All the best


christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 251
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: $rpacor $maxcor $ricore questions
« Reply #1 on: October 07, 2016, 01:59:12 pm »
All three of them refer exclusively the RAM.

$rpacor is special for the rirpa, $ricore for ridft, rdgrad, escf?, egrad?, rirpa?, aoforce? and might be replaced in the future by $maxcor

$maxcor is the orginal variable for the max. amount of RAM used by the TURBOMOLE programs for dynamic memory allocation in the Fortran code.
Note that on top of that the binaries need some memory themself and on Linux the MKL library, which is used for the linear algebra stuff, allocates
internally some extra RAM and the operating system needs some RAM to buffer I/O.
As a rule of thumb, we recommend to set $maxcor to about 66 - 75% of the physically available RAM, if you don't have a better estimate.

ricc2, ccsdf12, pnoccsd, mpgrad and mpshift use exclusively $maxcor. They ignore $rpacor and $ricore.

Christof