TCE - Problems with GA


Clicked A Few Times
Dear community and dear developers,

i have severe problems getting CC jobs running by using the TCE. I have created a minimal 4 hydrogen atom example that still doesn't work. Could you have a look at my in and outputs and give me a hint what I am missing ?

The computer is a 48 core shared memory machine ... so in principle the easiest option possible.

The non-TCE CCSD calculation works fine. Your help is greatly appreciated, as I need to run CC2 calculations for a quite big system.

best regards
Alex


title "test"
start test
memory stack 500 mb heap 100 mb global 600 mb
scratch_dir ./SCRATCHDIR
geometry print xyz 
  symmetry c1
H  4.18909 -0.161986 -0.00624883
H  2.00446507 -0.00532947  0.0000 
H  9.18909 -0.161986 -0.00624883
H  8.00446507 -0.00532947  0.0000 
end
basis spherical
 * library def2-TZVPP file /programme/basis/NWCHEM-BASIS-LIB/def2-tzvpp
end
scf
  singlet 
  rhf
  thresh 1.e-8
  maxiter 200
end
tce
  ccsd
  io fortran  
end
task tce energy


and

tce
  ccsd
  io replicated
end


the error messages are for the former and latter cases

Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation.

Backtrace for this error:

Backtrace for this error:
#0  #0  0x7F255C68B6F7
0x7FC36FCAB6F7
#1  #1  0x7FC36FCABCC4
#2  0x7F255C68BCC4
0x7FC36F1C544F
#2  0x7F255BBA544F
#0  0x7F0C20A936F7
#1  0x7F0C20A93CC4
#2  0x7F0C1FFAD44F
#0  0x7F49F3D236F7
#1  0x7F49F3D23CC4
#2  0x7F49F323D44F
#3  0x2AA0A11 in wnga_lock
#3  0x2AA0A11 in wnga_lock
#4  0x1537054 in createfile_
#4  0x1537054 in createfile_
#5  0x153D9CF in sf_test_
#5  0x153D9CF in sf_test_
#3  0x2AA0A11 in wnga_lock
#4  0x1537054 in createfile_
#5  0x153D9CF in sf_test_
#3  0x2AA0A11 in wnga_lock
#4  0x1537054 in createfile_
#5  0x153D9CF in sf_test_
#6  0x14E9BCE in tce_energy_ at tce_energy.F:759


and



 -----------------------------------------------------------------
 Iter          Residuum       Correlation     Cpu    Wall    V2*C2
 -----------------------------------------------------------------
0:nga_get_common:nga_get_common: ARRAY NOT ACTIVE:Received an Error in Communication
2:nga_get_common:nga_get_common: ARRAY NOT ACTIVE:Received an Error in Communication
application called MPI_Abort(comm=0x84000002, 7) - process 2
3:nga_get_common:nga_get_common: ARRAY NOT ACTIVE:Received an Error in Communication
application called MPI_Abort(comm=0x84000002, 7) - process 3
application called MPI_Abort(comm=0x84000004, 7) - process 0
1:nga_get_common:nga_get_common: ARRAY NOT ACTIVE:Received an Error in Communication
application called MPI_Abort(comm=0x84000002, 7) - process 1




Clicked A Few Times
Please do not use the fortran77 option in big calculation.
Instead, you should be using GA. Please send me your input file.
Best,
Karol

Clicked A Few Times
Thanks for answering.
here you find the respective in- and outputs of my tests

https://www.dropbox.com/s/nk08gbc1imp6rfu/IO-TCE-test_A_Kulesza.zip

The normal TCE default-GA run works for the minimal example, but my available memory is not sufficient for my real problem.

Its a 20 atom Fe-containing complex with TZVPP basis set

Best
Alex

Gets Around
Quote:CoolEsza Mar 13th 8:45 am
Thanks for answering.
here you find the respective in- and outputs of my tests

https://www.dropbox.com/s/nk08gbc1imp6rfu/IO-TCE-test_A_Kulesza.zip

The normal TCE default-GA run works for the minimal example, but my available memory is not sufficient for my real problem.

Its a 20 atom Fe-containing complex with TZVPP basis set

Best
Alex


You will likely need a bigger computer or a different code that does RI-CC2 or some other memory-reducing algorithm.


Forum >> NWChem's corner >> Running NWChem