Try to run the code interactively, or on a single processor: this should tell you something more on why the code crashes. Also try a smaller number of k-points. Note that by default all Kohn-Sham orbitals are stored in memory.
Paolo On Mon, Feb 25, 2019 at 1:31 AM Bharat Thapa <[email protected]> wrote: > Dear developers, > > I am running a scf calculation to test the cutoff convergence with respect > to the total energy. My plan is then to test k points and lattice constant > convergence test and then doing vc-relax. The system is Hf7O8N4. However, > surprising, it stops after printing: > > k( 763) = ( 0.4375000 -0.4375000 -0.3125000), wk = > 0.0026042 > k( 764) = ( -0.4375000 0.0000000 -0.4375000), wk = 0.0026042 > k( 765) = ( 0.4375000 -0.0000000 -0.4375000), wk = 0.0026042 > k( 766) = ( 0.0000000 -0.4375000 0.4375000), wk = 0.0026042 > k( 767) = ( 0.0000000 0.4375000 0.4375000), wk = 0.0026042 > k( 768) = ( 0.4375000 -0.4375000 -0.4375000), wk = 0.0026042 > > Dense grid: 1146249 G-vectors FFT dimensions: ( 144, 144, 135) > > Smooth grid: 220527 G-vectors FFT dimensions: ( 81, 81, 75) > > Largest allocated arrays est. size (Mb) dimensions > Kohn-Sham Wavefunctions 1.19 Mb ( 457, 170) > NL pseudopotentials 1.55 Mb ( 457, 222) > Each V/rho on FFT grid 0.95 Mb ( 62208) > Each G-vector array 0.14 Mb ( 17913) > G-vector shells 0.06 Mb ( 8140) > Largest temporary arrays est. size (Mb) dimensions > Auxiliary wavefunctions 4.74 Mb ( 457, 680) > Each subspace H/S matrix 0.28 Mb ( 136, 136) > Each <psi_i|beta_j> matrix 0.58 Mb ( 222, 170) > Arrays for rho mixing 9.49 Mb ( 62208, 10) > > Initial potential from superposition of free atoms > > starting charge 151.98640, renormalised to 152.00000 > Starting wfc are 118 randomized atomic wfcs + 52 random wfc > > 1763,6 Bot > > I am not sure what happened and why it stopped without any error message. > What should I do? I am running this parallel on 64 cores. I have never > found such strange situation before and was able to solve it via googling > but this time I am helpless. > > Please help me! > *Bharat Thapa* > _______________________________________________ > users mailing list > [email protected] > https://lists.quantum-espresso.org/mailman/listinfo/users -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
_______________________________________________ users mailing list [email protected] https://lists.quantum-espresso.org/mailman/listinfo/users
