[Pw_forum] pw.x stuck

Yongsheng Zhang yshzhang88 at yahoo.com
Thu Jun 2 22:12:41 CEST 2011


Firstly, this system was running throughout using previous PWSCF version, v4.2.1. 

Further tests with v4.3 using different CPUs, pw.x is stuck at different places. 

(1) 1CUP: after one static iteration, it leaves error message,
     WARNING: integrated charge=     0.00000000, expected=    64.00000000

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     from electrons : error #         1
     charge is wrong: smearing is needed
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

(2) 4 CPUs: the system did relaxed. But when it run the "final scf", it is stuck at, 
    A final scf calculation at the relaxed structure.

     The G-vectors are recalculated.

(3) 8 CPUs: after one static iteration, it is stuck at, 

        k(    1) = (   0.1249057   0.1249057   0.1531359), wk =   0.2500000
        k(    2) = (   0.1249057   0.3747172   0.1531359), wk =   0.5000000
        k(    3) = (   0.3747172   0.3747172   0.1531359), wk =   0.2500000
        k(    4) = (  -0.1249057   0.1249057  -0.1531359), wk =   0.2500000
        k(    5) = (  -0.1249057   0.3747172  -0.1531359), wk =   0.5000000
        k(    6) = (  -0.3747172   0.3747172  -0.1531359), wk =   0.2500000

I am kind of pretty sure that there is something wrong in the compiling the code. Then, I did my best to change the make.sys file and recompile the code. The best results I can get so far is: pw.x do relax the system, but when it is trying to run "A final scf calculation at the relaxed structure", it stops at, 
     G cutoff = 2508.9230  ( 852531 G-vectors)     FFT grid: (108,108,180)
     G cutoff = 1003.5692  ( 215507 G-vectors)  smooth grid: ( 64, 64,108)

     Largest allocated arrays     est. size (Mb)     dimensions
        Kohn-Sham Wavefunctions         1.65 Mb     (   3377,  32)
        NL pseudopotentials             5.77 Mb     (   3377, 112)
        Each V/rho on FFT grid          4.09 Mb     ( 268272)
        Each G-vector array             0.81 Mb     ( 106570)
        G-vector shells                 0.36 Mb     (  47252)
     Largest temporary arrays     est. size (Mb)     dimensions
        Auxiliary wavefunctions         6.60 Mb     (   3377, 128)
        Each subspace H/S matrix        0.25 Mb     (    128, 128)
        Each <psi_i|beta_j> matrix      0.05 Mb     (    112,  32)
        Arrays for rho mixing          32.75 Mb     ( 268272,   8)


And give the following error message:
forrtl: severe (41): insufficient virtual memory
Image              PC                Routine            Line        Source
pw.x               0000000001124F0D  Unknown               Unknown  Unknown
pw.x               0000000001123A15  Unknown               Unknown  Unknown
pw.x               00000000010C2C39  Unknown               Unknown  Unknown
pw.x               000000000105885D  Unknown               Unknown  Unknown
pw.x               000000000109736E  Unknown               Unknown  Unknown
pw.x               000000000053EEBF  allocate_nlpot_            68  allocate_nlpot.f90
pw.x               0000000000454D76  init_run_                  60  init_run.f90
pw.x               00000000004716B7  move_ions_                290  move_ions.f90
pw.x               000000000044AA55  MAIN__                    173  pwscf.f90
pw.x               000000000044A60C  Unknown               Unknown  Unknown
libc.so.6          00002AE8FFD68974  Unknown               Unknown  Unknown
pw.x               000000000044A519  Unknown               Unknown  Unknown


I am wondering that why the "insufficient virtual memory" happens at the last step. If there is no sufficient virtual memory, the code should not even run at the first iteration. While, I collect the relaxed structure and re-run pw.x, it works without any problem. 

Thanks
Yongsheng


--- On Thu, 6/2/11, Paolo Giannozzi <giannozz at democritos.it> wrote:

> From: Paolo Giannozzi <giannozz at democritos.it>
> Subject: Re: [Pw_forum] pw.x stuck
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Date: Thursday, June 2, 2011, 1:45 AM
> 
> On Jun 1, 2011, at 19:05 , Jack London wrote:
> 
> > Yes, I have tried many times
> 
> many times in the same conditions is better than nothing,
> but you should try in different conditions as well.
> 
> P.
> ---
> Paolo Giannozzi, Dept of
> Chemistry&Physics&Environment,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
> 
> 
> 
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 



More information about the users mailing list