If you have problems during the execution of MRCC, please attach the output with an adequate description of your case as well as the followings:
  • the way mrcc was invoked
  • the way build.mrcc was invoked
  • the output of build.mrcc
  • compiler version (for example: ifort -V, gfortran -v)
  • blas/lapack versions
  • as well as gcc and glibc versions

This information really helps us during troubleshooting :)

The mystery of the change in memory requirements

  • Nike
  • Topic Author
  • Offline
  • Premium Member
  • Premium Member
More
7 years 3 weeks ago #449 by Nike
The mystery of the change in memory requirements was created by Nike
Greetings!
In my calculation below, first it says that 122631 MB (123GB) is the minimum needed, but then it re-executes Goldstone, and xmrcc, and then suddenly it says that the minimum RAM required is 240GB, which is ALMOST double, but not exactly double the original amount reportedly needed.
Code:
CC(3)(4) calculation Allocation of****** Mbytes of memory... Number of spinorbitals: 756 Number of alpha electrons: 8 Number of beta electrons: 8 Spin multiplicity: 1 z-component of spin: 0.0 Spatial symmetry: 1 Convergence criterion: 1.0E-07 Construction of occupation graphs... Number of 0 -fold excitations: 1 Number of 1 -fold excitations: 888 Number of 2 -fold excitations: 1600580 Number of 3 -fold excitations: 1533088192 Total number of configurations: 1534689661 Calculation of coupling coefficients... Length of intermediate file (Mbytes): 32072.1 ====================================================================== Spin case 1 Alpha: 0 Beta: 4 Number of excitations: 6720740812 Spin case 2 Alpha: 1 Beta: 3 Number of excitations: 173490018688 Spin case 3 Alpha: 2 Beta: 2 Number of excitations: 456717146288 Number of 4 -fold excitations: 817138665288 Memory requirements /Mbyte/: Minimal Optimal Real*8: 120561.4595 230562.1549 Integer: 2069.9997 Total: 122631.4591 232632.1545 ************************ 2017-10-27 10:24:46 ************************* Executing goldstone... Generation of CC equations in terms of H and T... Generation of antisymmetrized Goldstone diagrams... Number of diagrams in T^1 equations: 15 Number of diagrams in T^2 equations: 38 Number of diagrams in T^3 equations: 49 Number of diagrams in T^4 equations: 5 Translation of diagrams to factorized equations... Optimizing intermediate calculation... Number of floating-point operations per iteration step: 1.0740E+16 Probable CPU time per iteration step (hours): 107401.43 Required memory (Mbytes): ********* Number of intermediates: 79 Number of intermediates to be stored: 32 Length of intermediate file (Mbytes): 42887.7 ************************ 2017-10-27 10:24:50 ************************* Executing xmrcc... ********************************************************************** CC(3)(4) calculation Allocation of****** Mbytes of memory... Number of spinorbitals: 756 Number of alpha electrons: 8 Number of beta electrons: 8 Spin multiplicity: 1 z-component of spin: 0.0 Spatial symmetry: 1 Convergence criterion: 1.0E-07 Construction of occupation graphs... Number of 0 -fold excitations: 1 Number of 1 -fold excitations: 888 Number of 2 -fold excitations: 1600580 Number of 3 -fold excitations: 1533088192 Total number of configurations: 1534689661 Calculation of coupling coefficients... Length of intermediate file (Mbytes): 32072.1 ====================================================================== Spin case 1 Alpha: 0 Beta: 4 Number of excitations: 6720740812 Spin case 2 Alpha: 1 Beta: 3 Number of excitations: 173490018688 Spin case 3 Alpha: 2 Beta: 2 Number of excitations: 456717146288 Number of 4 -fold excitations: 817138665288 Memory requirements /Mbyte/: Minimal Optimal Real*8: 238711.7790 256640.9468 Integer: 2069.9997 Total: 240781.7787 258710.9465 ************************ 2017-10-27 10:41:46 ************************* Executing goldstone... Generation of MRCC equations in terms of H and T... Generation of antisymmetrized Goldstone diagrams... Number of diagrams in T^1 equations: 15 Number of diagrams in T^2 equations: 38 Number of diagrams in T^3 equations: 49 Number of diagrams in T^4 equations: 5 Translation of diagrams to factorized equations... Number of restricted diagrams in T^1 equations: 78 Number of restricted diagrams in T^2 equations: 344 Number of restricted diagrams in T^3 equations: 822 Number of restricted diagrams in T^4 equations: 86 Optimizing intermediate calculation... Number of floating-point operations per iteration step: 1.0767E+16 Probable CPU time per iteration step (hours): 107668.68 Required memory (Mbytes): 8088493.9 Number of intermediates: 534 Number of intermediates to be stored: 217 Length of intermediate file (Mbytes): 65406.0 ************************ 2017-10-27 10:42:01 ************************* Executing xmrcc... ********************************************************************** MRCC( 4 ) calculation Allocation of****** Mbytes of memory... Number of spinorbitals: 756 Number of alpha electrons: 8 Number of beta electrons: 8 Number of active particles: 370 Number of active holes: 0 Spin multiplicity: 1 z-component of spin: 0.0 Spatial symmetry: 1 Convergence criterion: 1.0E-07 Construction of occupation graphs... Number of 0 -fold excitations: 1 Number of 1 -fold excitations: 888 Number of 2 -fold excitations: 1600580 Number of 3 -fold excitations: 1533088192 Total number of configurations: 1534689661 Calculation of coupling coefficients... Request for 18066976 words of memory failed. Fatal error in xmrcc. Program will stop.

For a long time I've been curious about what it's doing when the memory requirement keeps changing. I understand that if I've not allocated enough RAM (in this case 123GB), then MRCC will try to do something to make the RAM requirement smaller, but the usual behavior is that the minimum RAM requirement goes down (or only slightly up) with each iteration of Goldstone, but in this case the RAM requirement almost doubled.

1) What's it doing that is making it require twice as much RAM?
2) Why make it require MORE memory, if the problem is that the required RAM is not available?
3) Usually MRCC keeps re-executing Goldstone and xmrcc until the RAM requirement is low enough, but this time it gave up after only two iterations, why is that?

With best wishes!!
Nike Dattani

Please Log in or Create an account to join the conversation.

  • Nike
  • Topic Author
  • Offline
  • Premium Member
  • Premium Member
More
7 years 3 weeks ago - 7 years 3 weeks ago #450 by Nike
Replied by Nike on topic The mystery of the change in memory requirements
Greetings!
I should have attached the input file. Here it is (the iface=cfour is just to read the fort.55 that MRCC already produced earlier).

The basis set aV5Z-EMSL is just the aug-cc-pV5Z found on EMSL basis set exchange, which might already be in MRCC, but I like to manually enter the basis set every time.
Code:
# TITLE basis=aV5Z-EMSL iface=cfour uncontract=off itol=15 calc=CCSDT(Q) mem=110GB core=3 cctol=7 ccmaxit=999 scfmaxit=9999 scftype=RHF scfiguess=ao rohftype=standard mult=1 charge=0 rest=2 unit=angs geom C O 1 R O 1 R 2 OCO R=1.163 OCO=180
Last edit: 7 years 3 weeks ago by Nike.

Please Log in or Create an account to join the conversation.

Time to create page: 0.038 seconds
Powered by Kunena Forum

MRCC Login