[CP2K:5528] The MEMARY consumption of RI_MP2 calculation of periodic system

hut... at chem.uzh.ch hut... at chem.uzh.ch
Fri Jul 18 08:11:32 UTC 2014


Hi

without knowing details about your system and basis sets, I would
estimate that you need about 10TB of memory (absolute minimum)
for the calculation.
For your setup with 128GB/node, it means that below 100 nodes
you will not be able to perform the calculation.

BTW this would be close to the largest MP2 calculation ever done
 
regards

Juerg Hutter
--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp... at googlegroups.com wrote: -----
To: cp... at googlegroups.com
From: Bin Gu 
Sent by: cp... at googlegroups.com
Date: 07/18/2014 12:42AM
Subject: [CP2K:5528] The MEMARY consumption of RI_MP2 calculation of  periodic system

Hi everyone: 

I am trying MP2 for a condensed system (21^3 AA cell with 1000  C H O atoms) with  xyz periodic conditions, on a cluster of 10 nodes with the memory of  128GB/node and 24 processors/node.   

I set the HF and MP2 section as: 

 &HF
        FRACTION 1.0
        &SCREENING
          EPS_SCHWARZ 5.0E-10
        &END
        &INTERACTION_POTENTIAL
          POTENTIAL_TYPE TRUNCATED
          CUTOFF_RADIUS  10.0
          T_C_G_DATA  ../t_c_g.dat
        &END
        &MEMORY
          MAX_MEMORY 750
        &END
 &END

 &WF_CORRELATION
           METHOD  RI_MP2_GPW
           MEMORY    1000  # the memory
           NUMBER_PROC 10
         &WFC_GPW
           CUTOFF      300
           REL_CUTOFF  50
           EPS_FILTER  1.0E-12
           EPS_GRID    1.0E-8
         &END
 &END

But the calculation always stopped at the output of: 

HFX_MEM_INFO| Est. max. program size before HFX [MB's]:     5xx MB

with an Error Messages of Out of Memory. 

For our system, the Memory available for every node is 128G/24=5.3 GB. 
In this way, it seems the setting of   HF : 750Mb, and 1GB for MP2 are reasonable. 

Is this right?  
How the memory consumption of MP2 is determined for the MPI implementation?
Is NUMBER_PROC 10 relevent ?  I set it as the total number of nodes is 10. 

Thanks very much for any information.

Bin Gu
  
  -- 
 You received this message because you are subscribed to the Google Groups "cp2k" group.
 To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.
 To post to this group, send email to cp... at googlegroups.com.
 Visit this group at http://groups.google.com/group/cp2k.
 For more options, visit https://groups.google.com/d/optout.




More information about the CP2K-user mailing list