[SOLVED] *** Error in MKSRT1 ***

chucksty · 2019-12-25 14:38:27

I am running a job using openMolcas and have been having this memory issue which increasing Molcas_mem doesn't seem to be fixing. The error message as below;

*** Error in MKSRT1 ***
Insufficient memory
nBin exceeds limits (nBin>mxBin)
nBin= 72779
mxBin= 2048
Increase MOLCAS_MEM and try again!
--- Stop Module: seward at Sat Dec 21 03:13:32 2019 /rc=_RC_MEMORY_ERROR_ ---

I have set MOLCAS_MEM = 64000, but that doesn't help. I have no idea how to increase mxBin or if it is even possible to do so without looking at the src. I have also tried to run pymolcas -setup and rest the default MOLCAS_MEM to 32000 but this doesn't fix anything. Any idea how to deal with this?

Thanks

David

Ignacio · 2019-12-27 12:24:47

The effect of increasing the memory should be to reduce nBin. Maybe you are not setting/exporting the value properly or something else is overriding it. Check the reported memory in the SEWARD header, e.g.:

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &SEWARD

                       only a single process is used, running in SERIAL mode
                       available to each process: 2.0 GB of memory, 8 threads
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

Such a high nBin number probably indicates a very large basis set or very little memory. You could try with RICD/Cholesky too.

chucksty · 2019-12-27 14:14:28

Thank you very much for your response, Ignacio. Indeed the basis set is large (ANO-RCC). Below is the &seward readout;

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &SEWARD

                                   only a single process is used
                       available to each process: 36 GB of memory, 1 thread?
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

               SEWARD will generate:
                  Multipole Moment integrals up to order  2
                  Kinetic Energy integrals
                  Nuclear Attraction integrals (point charge)
                  One-Electron Hamiltonian integrals
                  Velocity integrals
                  Relativistic Douglas-Kroll-Hess integrals:
                    - Parametrization         : EXP
                    - DKH order of Hamiltonian: 2
                    - DKH order of Properties : 0
                         - multipole moment operators
                         - electric potential operators
                         - contact operators
                  Atomic mean-field integrals
                  Two-Electron Repulsion integrals

                   Integrals are discarded if absolute value <: 0.10E-13
                   Integral cutoff threshold is set to       <: 0.10E-15

            Nuclear Potential Energy           7289.31214529 au


      Basis set specifications :
      Symmetry species         a
      Basis functions         5684


  *** Error in MKSRT1 ***
  Insufficient memory
  nBin exceeds limits (nBin>mxBin)

 nBin=               129255
 mxBin=                 2048

 Increase MOLCAS_MEM and try again!

--- Stop Module: seward at Fri Dec 27 12:46:48 2019 /rc=_RC_MEMORY_ERROR_ ---
*** files: xmldump
    saved to directory /home/dci24/molcas_calculation/Dy-Cp/RASSCF/dy-Cp2-Cl-60k/"PWD"
--- Module seward spent 1 hour 6 minutes 35 seconds ---

    Timing: Wall=4001.57 User=3955.89 System=31.88

I have also tried this with 36 GB of memory. This doesn't fix the problem.

The settings I have in my submission script is below;

######## Job settings ###########
export MOLCAS_MEM=36000
export SUBMIT=/home/dci24/molcas_calculation/
export Project=dy-complex-rassscf
export MOLCAS_NPROCS=48
export MOLCAS_PRINT=NORMAL #####Other options are Silent (0), Terse, Normal, Verbose, Debug, Insane (5)
export MOLCAS_DISK=48000

is their something I am doing wrong?
Thanks

David

Ignacio · 2019-12-28 09:45:37

It seems correct (the header says 36 GB). I'm afraid 5700 basis functions are a lot, you'll probably need more memory to handle them (and massive disk space, and time).

My recommendation is to use RICD, though it will still be a challenging calculation.

chucksty · 2020-03-09 12:07:40

Just to mention that the use of RICD in &gateway or cholesky=High in &seward resolved this problem I was having here. Thanks admin.

Chucksty

Molcas Forum

Announcement

#1 2019-12-25 14:38:27