6
$\begingroup$

I performed a vc-relax calculation with hybrid functional on a 2 atoms material using the command mpirun -np 16 --oversubscribe 'directory/qe/pw.x' -in vc-relax.in > vc-relax.out. It's been 6 hours now since the code is running and still didn't finish yet.

I want to know if my simulation is working properly or there is something wrong with it. Is there a way to know how much time a simulation will take?

I extracted two separated parts of my output file here.

-Part 1:

Dense  grid:    50701 G-vectors     FFT dimensions: (  54,  54,  54)

     Dynamical RAM for                 wfc:       0.11 MB

     Dynamical RAM for     wfc (w. buffer):       1.84 MB

     Dynamical RAM for                 EXX:     380.85 MB

     Dynamical RAM for           str. fact:       0.10 MB

     Dynamical RAM for           local pot:       0.00 MB

     Dynamical RAM for          nlocal pot:       0.22 MB

     Dynamical RAM for                qrad:       6.86 MB

     Dynamical RAM for          rho,v,vnew:       0.41 MB

     Dynamical RAM for               rhoin:       0.14 MB

     Dynamical RAM for            rho*nmix:       0.77 MB

     Dynamical RAM for           G-vectors:       0.21 MB

     Dynamical RAM for          h,s,v(r/c):       0.06 MB

     Dynamical RAM for          <psi|beta>:       0.01 MB

     Dynamical RAM for                 psi:       0.22 MB

     Dynamical RAM for                hpsi:       0.22 MB

     Dynamical RAM for      wfcinit/wfcrot:       0.24 MB

     Estimated static dynamical RAM per process >     391.22 MB

     Estimated max dynamical RAM per process >     392.49 MB

     Estimated total dynamical RAM >       6.13 GB

     Initial potential from superposition of free atoms

     starting charge      35.9998, renormalised to      36.0000
     Starting wfcs are   19 randomized atomic wfcs

     total cpu time spent up to now is        0.6 secs

     Self-consistent Calculation

     iteration #  1     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            39 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12030 MiB available memory on the node where the printing process lives
------------------
     ethr =  1.00E-02,  avg # of iterations =  2.0

     Threshold (ethr) on eigenvalues was too large:
     Diagonalizing with lowered threshold

     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            40 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12036 MiB available memory on the node where the printing process lives
------------------
     ethr =  7.86E-04,  avg # of iterations =  1.1

     total cpu time spent up to now is        1.6 secs

     total energy              =    -567.82636984 Ry
     estimated scf accuracy    <       0.28065174 Ry

     iteration #  2     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            41 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12015 MiB available memory on the node where the printing process lives
------------------
     ethr =  7.80E-04,  avg # of iterations =  1.1

     total cpu time spent up to now is        2.2 secs

     total energy              =    -567.76398782 Ry
     estimated scf accuracy    <       0.13999925 Ry

     iteration #  3     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            41 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12044 MiB available memory on the node where the printing process lives
------------------
     ethr =  3.89E-04,  avg # of iterations =  2.1

     total cpu time spent up to now is        2.9 secs

     total energy              =    -567.79592707 Ry
     estimated scf accuracy    <       0.00815880 Ry

     iteration #  4     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            41 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12069 MiB available memory on the node where the printing process lives
------------------
     ethr =  2.27E-05,  avg # of iterations =  2.1

     total cpu time spent up to now is        3.4 secs

     total energy              =    -567.79503173 Ry
     estimated scf accuracy    <       0.00477159 Ry

     iteration #  5     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            41 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12069 MiB available memory on the node where the printing process lives
------------------
     ethr =  1.33E-05,  avg # of iterations =  2.2

     total cpu time spent up to now is        4.0 secs

     total energy              =    -567.79436235 Ry
     estimated scf accuracy    <       0.00667744 Ry

     iteration #  6     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12091 MiB available memory on the node where the printing process lives
------------------
     ethr =  1.33E-05,  avg # of iterations =  2.1

     total cpu time spent up to now is        4.6 secs

     total energy              =    -567.79518241 Ry
     estimated scf accuracy    <       0.00031137 Ry

     iteration #  7     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12089 MiB available memory on the node where the printing process lives
------------------
     ethr =  8.65E-07,  avg # of iterations =  2.2

     total cpu time spent up to now is        5.1 secs

     total energy              =    -567.79532210 Ry
     estimated scf accuracy    <       0.00014763 Ry

     iteration #  8     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12088 MiB available memory on the node where the printing process lives
------------------
     ethr =  4.10E-07,  avg # of iterations =  3.3

     total cpu time spent up to now is        5.7 secs

     total energy              =    -567.79490852 Ry
     estimated scf accuracy    <       0.00143778 Ry

     iteration #  9     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12086 MiB available memory on the node where the printing process lives
------------------
     ethr =  4.10E-07,  avg # of iterations =  3.1

     total cpu time spent up to now is        6.6 secs

     total energy              =    -567.79517589 Ry
     estimated scf accuracy    <       0.00002797 Ry

     iteration # 10     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12054 MiB available memory on the node where the printing process lives
------------------
     ethr =  7.77E-08,  avg # of iterations =  3.2

     total cpu time spent up to now is        7.3 secs

     total energy              =    -567.79517480 Ry
     estimated scf accuracy    <       0.00006803 Ry

     iteration # 11     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12053 MiB available memory on the node where the printing process lives
------------------
     ethr =  7.77E-08,  avg # of iterations =  2.0

     total cpu time spent up to now is        8.1 secs

     total energy              =    -567.79516053 Ry
     estimated scf accuracy    <       0.00002961 Ry

     iteration # 12     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12085 MiB available memory on the node where the printing process lives
------------------
     ethr =  7.77E-08,  avg # of iterations =  3.0

     total cpu time spent up to now is        8.8 secs

     total energy              =    -567.79517153 Ry
     estimated scf accuracy    <       0.00000876 Ry

     iteration # 13     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12084 MiB available memory on the node where the printing process lives
------------------
     ethr =  2.43E-08,  avg # of iterations =  2.0

     total cpu time spent up to now is        9.3 secs

     total energy              =    -567.79517144 Ry
     estimated scf accuracy    <       0.00000216 Ry

     iteration # 14     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12084 MiB available memory on the node where the printing process lives
------------------
     ethr =  6.00E-09,  avg # of iterations =  2.1

     total cpu time spent up to now is        9.9 secs

     total energy              =    -567.79517161 Ry
     estimated scf accuracy    <       0.00000034 Ry

     iteration # 15     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12084 MiB available memory on the node where the printing process lives
------------------
     ethr =  9.49E-10,  avg # of iterations =  1.3

     total cpu time spent up to now is       10.6 secs

     total energy              =    -567.79517158 Ry
     estimated scf accuracy    <       0.00000025 Ry

     iteration # 16     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12084 MiB available memory on the node where the printing process lives
------------------
     ethr =  6.98E-10,  avg # of iterations =  1.8

     total cpu time spent up to now is       11.1 secs

     total energy              =    -567.79517154 Ry
     estimated scf accuracy    <       0.00000011 Ry

     iteration # 17     ecut=   100.00 Ry     beta= 0.10
     Davidson diagonalization with overlap

---- Real-time Memory Report at c_bands before calling an iterative solver
            42 MiB given to the printing process from OS
             0 MiB allocation reported by mallinfo(arena+hblkhd)
         12084 MiB available memory on the node where the printing process lives
------------------
     ethr =  2.92E-10,  avg # of iterations =  2.5

     total cpu time spent up to now is       12.3 secs

-Part 2:


     Using ACE for calculation of exact exchange

!!   total energy              =    -567.16587258 Ry
     Harris-Foulkes estimate   =    -567.16587258 Ry
     est. exchange err (dexx)  =          2.5E-09 Ry
     - averaged Fock potential =      24.73517817 Ry
     + Fock energy (ACE)       =     -12.36762662 Ry

     EXX self-consistency reached

     Forces acting on atoms (cartesian axes, Ry/au):

     atom    1 type  2   force =     0.00000000    0.00000000    0.00000000
     atom    2 type  1   force =     0.00000000    0.00000000    0.00000000
     The non-local contrib.  to forces
     atom    1 type  2   force =     0.00000000    0.00000000    0.00000000
     atom    2 type  1   force =     0.00000000    0.00000000    0.00000000
     The ionic contribution  to forces
     atom    1 type  2   force =    -0.00000000   -0.00000000   -0.00000000
     atom    2 type  1   force =    -0.00000000   -0.00000000    0.00000000
     The local contribution  to forces
     atom    1 type  2   force =    -0.00000000    0.00000000   -0.00000000
     atom    2 type  1   force =    -0.00000000   -0.00000000   -0.00000000
     The core correction contribution to forces
     atom    1 type  2   force =     0.00000000    0.00000000    0.00000000
     atom    2 type  1   force =     0.00000000    0.00000000    0.00000000
     The Hubbard contrib.    to forces
     atom    1 type  2   force =     0.00000000    0.00000000    0.00000000
     atom    2 type  1   force =     0.00000000    0.00000000    0.00000000
     The SCF correction term to forces
     atom    1 type  2   force =     0.00000000    0.00000000   -0.00000000
     atom    2 type  1   force =     0.00000000   -0.00000000   -0.00000000

     Total force =     0.000000     Total SCF correction =     0.000000


     Computing stress (Cartesian axis) and pressure

          total   stress  (Ry/bohr**3)                   (kbar)     P=      -27.97
  -0.00019011   0.00000000   0.00000000          -27.97        0.00        0.00
   0.00000000  -0.00019011  -0.00000000            0.00      -27.97       -0.00
   0.00000000  -0.00000000  -0.00019011            0.00       -0.00      -27.97

     kinetic stress (kbar)  68348.60     -0.00     -0.00
                               -0.00  68348.60      0.00
                               -0.00      0.00  68348.60

     local   stress (kbar) -12153.28     -0.00     -0.00
                               -0.00 -12153.28     -0.00
                               -0.00     -0.00 -12153.28

     nonloc. stress (kbar) -38015.15      0.00     -0.00
                               -0.00 -38015.15      0.00
                                0.00      0.00 -38015.15

     hartree stress (kbar)  28570.52      0.00     -0.00
                                0.00  28570.52     -0.00
                               -0.00     -0.00  28570.52

     exc-cor stress (kbar)  -5191.68      0.00      0.00
                                0.00  -5191.68     -0.00
                                0.00     -0.00  -5191.68

     corecor stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00

     ewald   stress (kbar) -39936.83      0.00      0.00
                                0.00 -39936.83     -0.00
                                0.00     -0.00 -39936.83

     hubbard stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00

     DFT-D   stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00

     XDM     stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00

     dft-nl  stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00

     TS-vdW  stress (kbar)      0.00      0.00      0.00
                                0.00      0.00      0.00
                                0.00      0.00      0.00


                                0.00      0.00      0.00
                                0.00      0.00      0.00


                                0.00      0.00      0.00

     EXX     stress (kbar)  -1650.14     -0.00     -0.00
                               -0.00  -1650.14      0.00
                               -0.00      0.00  -1650.14



     BFGS Geometry Optimization
     Energy error            =      0.0E+00 Ry
     Gradient error          =      0.0E+00 Ry/Bohr
     Cell gradient error     =      2.8E+01 kbar

     number of scf cycles    =   1
     number of bfgs steps    =   0

     enthalpy           new  =    -567.1658725777 Ry

     new trust radius        =       0.0184108950 bohr
     Message from routine volume:
     axis vectors are left-handed
     new conv_thr            =       0.0000000100 Ry

     new unit-cell volume =    369.83860 a.u.^3 (    54.80443 Ang^3 )

-This is a part extracted from my input file:

&CONTROL
 ...
  pseudo_dir='.',
  verbosity='high',
  tprnfor=.true.,
  tstress=.true.,
  etot_conv_thr =1e-8
  forc_conv_thr =1e-7

/
&SYSTEM
  ....
  input_dft='pbe0',
  nqx1 = 6, nqx2 =6, nqx3 =6,
  x_gamma_extrapolation = .true.
  exxdiv_treatment = 'gygi-baldereschi'
  exx_fraction = 0.25
  !ecutfock=100

/
&ELECTRONS
  conv_thr=1d-08,
  mixing_beta=0.1d0,
  electron_maxstep = 3000
  mixing_mode = 'plain' 
/

&IONS
  ion_dynamics='bfgs'
/
&CELL
  cell_dynamics='bfgs'
  press=0.d0,
  press_conv_thr=0.2d0
/
...

PS: The ... means that there is text there also

$\endgroup$
4
  • $\begingroup$ Could you provide the output file as a code block? Even if the calculation is not finished, you can open the output file. If the calculation is running for way above the expected time, there might be something else that might be traced back from the output file. Aside from what is suggested in the answer, maybe check the 'dynamic ram usage'/'total amount of ram needed' quantity and see if you have that much ram. Sometimes if the number of ram needed is more than your system has, the calculation runs forever. $\endgroup$ Commented Apr 5, 2023 at 17:34
  • $\begingroup$ @AbdulMuhaymin I edited my post can you check it again please. ram is used so many times in the ouptut file can you tell me which lines indicate the number of needed ram and the one that the system have. $\endgroup$
    – Camilla
    Commented Apr 5, 2023 at 21:09
  • 1
    $\begingroup$ Can you include your input file? What are the values that you have set for convergence criteria? $\endgroup$ Commented Apr 6, 2023 at 3:20
  • $\begingroup$ @HemanthHaridas I edited my post again and added a part of my input file $\endgroup$
    – Camilla
    Commented Apr 6, 2023 at 9:25

1 Answer 1

5
$\begingroup$

TL;DR

There were a few issues with the input file and the way OP was running the code.

Longer Answer

  1. OP had been using --oversubscribe flag hoping to use all the 16 virtual processors available on the laptop. Instead, OP needed to use
mpiexec -np 8 pw.x -in inputfile.in > outputfile.out

This ensures that all the 8 available physical cores are used for the calculation, and you are not throwing more that what the cores can handle at it. This point was reiterated by Phil Hasnip in this comment.

  1. A quick look at the input file (which was shared in chat) showed that there were a few issues with the input file, particularly with the unit cell vectors and the atomic positions. To check that, I downloaded the cif file for CuI from here and started a vc-relax run. I was able to get the calculation to complete 2 geometry optimization steps in 15 minutes (which is an improvement from what OP reported).
$\endgroup$
10
  • 9
    $\begingroup$ @Camilla oversubscribing doesn't mean using all the processors, it means trying to use more than the number of processors. It's an option for testing, and because modern processors do have some limited ability to run two things at once ("hyperthreading"), but in my experience trying it for DFT calculations always slows the simulation down. I recommend only using up to the actual number of cores on your machine, and ignore any hyperthreading completely. $\endgroup$ Commented Apr 6, 2023 at 0:16
  • 1
    $\begingroup$ @Camilla I have pinged you in this chat room Please join here, I will try explaining the issues. $\endgroup$ Commented Apr 6, 2023 at 10:23
  • 1
    $\begingroup$ Prof. Dattani, it was a mistake from my part. Thanks for the shifting the messages to proper room $\endgroup$ Commented Apr 6, 2023 at 15:39
  • 4
    $\begingroup$ @Camilla it's because your machine doesn't have 16 processors, it only has 8. The reason it appears as if it has 16 is because each processor can try to run two tasks at once ("hyperthreading"), but this is designed for when the two tasks are either not intensive or use different parts of the processor. It would be very unusual for a materials modelling program to benefit from this, it would almost always slow down. $\endgroup$ Commented Apr 6, 2023 at 23:33
  • 1
    $\begingroup$ @Camilla, Did the answer solve the issue that you had with the QE run? If yes, please indicate so in the answer. Thanks. $\endgroup$ Commented Apr 24, 2023 at 15:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .