[CPMD-list] cpmd v3.9.1 OpenMP version buggy

Axel Kohlmeyer axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Fri Jun 4 10:22:32 CEST 2004


On Thu, 3 Jun 2004, Alessandro Curioni wrote:

AC> 
AC> 
AC> 
AC> 
AC> Bernd,
AC> 
AC> thank you for your report, you have found the first problem of CPMD v3.9.
AC> 
AC> Please comment/delete  the OMP directives in the file  rnlin.F at  lines
AC> 212 and 288 and recompile the program.

hi everybody,

that is not the only OpenMP problem. i just tried the corrected version
with the PGI compilers (version 5.1 on x86_64, and version 3.2 on ix86)
and it seems that even compiling for OpenMP without using it triggers
a problem. for example on the x86_64 machines,

with 'OMP_NUM_THREADS=1 ./cpmd.x inp.wf' i get:

 ****************************************************************


 DEGREES OF FREEDOM FOR SYSTEM:                               186

 CPU TIME FOR WAVEFUNCTION INITIALIZATION:           3.64 SECONDS
 ***    RWFOPT| THE NEW SIZE OF THE PROGRAM IS   16320/ 113476 kBYTES ***
  EWALD| SUM IN REAL SPACE OVER                     5* 5* 5 CELLS
 UNIV| THE LEADING MINOR OF ORDER    1 IS NOT POSITIVE DEFINITE, 
 UNIV| AND THE FACTORIZATION COULD NOT BE COMPLETED.
 STOPGM! STACK OF MAIN CALLS:
 STOPGM! CALL        RGS


 PROGRAM STOPS IN SUBROUTINE UINV| ILLEGAL RESULTS DPOTRF
  999


with 'OMP_NUM_THREADS=2 ./cpmd.x inp.wf' i get:


 GENERATE ATOMIC BASIS SET
     Si        SLATER ORBITALS
        3S        ALPHA=   1.6344      OCCUPATION= 2.00
        3P        ALPHA=   1.4284      OCCUPATION= 2.00


 INITIALIZATION TIME:                                0.10 SECONDS

 ***    WFOPTS| THE NEW SIZE OF THE PROGRAM IS    4764/ 137712 kBYTES ***
 ***     PHFAC| THE NEW SIZE OF THE PROGRAM IS    4944/ 146256 kBYTES ***
 ***    ATOMWF| THE NEW SIZE OF THE PROGRAM IS    7448/ 156512 kBYTES ***
  ATRHO| CHARGE(R-SPACE):  252.000000      (G-SPACE):  252.000000

 DSYGVX| INFO=  251
 DSYGVX| FAILED TO CONVERGE WHEN DIAGONALIZING A
 STOPGM! STACK OF MAIN CALLS:
 STOPGM! CALL     ATOMWF


 PROGRAM STOPS IN SUBROUTINE DSYGVX| FAILED TO DIAGONALIZE
  999


with the 'normal' binary (from the _exact_ same sources) 
it works fine. 


and while on the subject of OpenMP. has anyone yet succeeded
to compile an OpenMP enabled CPMD binay with the intel fortran compiler?
i'd really like to hear about it. in my tests, the speed gain from
using an OpenMP enabled binary on dual-xeon or dual-athlon machines 
was always eaten up by the fact that the intel compiler produces
faster code than the pgi compiler (where OpenMP used to work).

best regards,
	axel kohlmeyer.

AC> 
AC> Best Regards,
AC> 
AC> 
AC> 
AC> 
AC> 
AC> Alessandro CURIONI, PhD
AC> Research Staff Member
AC> Computational Biochemistry and Material Science group
AC> IBM Research Division - Zurich Research Laboratory
AC> Saumerstrasse 4
AC> 8003 Rueschlikon - Switzerland
AC> e-mail: cur at zurich.ibm.com
AC> www:    www.zurich.ibm.com
AC> Tel: +41-1-7248633
AC> Fax: +41-1-7248958
AC> 
AC> _______________________________________________
AC> CPMD-list mailing list
AC> CPMD-list at cpmd.org
AC> http://cpmd.org/mailman/listinfo/cpmd-list
AC> 
AC> 

-- 


=======================================================================
Dr. Axel Kohlmeyer                        e-mail: axel.kohlmeyer at rub.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================

-------------- next part --------------
&CPMD
  OPTIMIZE WAVEFUNCTION
  MAXSTEP
    50
  ODIIS
    15
  TIMESTEP
    15.0
  MAXCPUTIME
   3600
  SPLINE POINTS
   1000
&END

&SYSTEM
  SYMMETRY
    1
  CELL
    20.5224280754 1.0 1.0 0.0 0.0 0.0
  CUTOFF
    10
  DUAL
     3
  TESR
    2 2 2
  SCALED
&END

&ATOMS
*SI_SGS   KLEINMAN-BYLANDER  RAGGIO=1.0
  LMAX=P LOC=P
      63
    0.000    0.250    0.250
    0.250    0.000    0.250
    0.250    0.250    0.000
    0.125    0.125    0.125
    0.125    0.375    0.375
    0.375    0.125    0.375
    0.375    0.375    0.125
    0.500    0.000    0.000
    0.500    0.250    0.250
    0.750    0.000    0.250
    0.750    0.250    0.000
    0.625    0.125    0.125
    0.625    0.375    0.375
    0.875    0.125    0.375
    0.875    0.375    0.125
    0.000    0.500    0.000
    0.000    0.750    0.250
    0.250    0.500    0.250
    0.250    0.750    0.000
    0.125    0.625    0.125
    0.125    0.875    0.375
    0.375    0.625    0.375
    0.375    0.875    0.125
    0.500    0.500    0.000
    0.500    0.750    0.250
    0.750    0.500    0.250
    0.750    0.750    0.000
    0.625    0.625    0.125
    0.625    0.875    0.375
    0.875    0.625    0.375
    0.875    0.875    0.125
    0.000    0.000    0.500
    0.000    0.250    0.750
    0.250    0.000    0.750
    0.250    0.250    0.500
    0.125    0.125    0.625
    0.125    0.375    0.875
    0.375    0.125    0.875
    0.375    0.375    0.625
    0.500    0.000    0.500
    0.500    0.250    0.750
    0.750    0.000    0.750
    0.750    0.250    0.500
    0.625    0.125    0.625
    0.625    0.375    0.875
    0.875    0.125    0.875
    0.875    0.375    0.625
    0.000    0.500    0.500
    0.000    0.750    0.750
    0.250    0.500    0.750
    0.250    0.750    0.500
    0.125    0.625    0.625
    0.125    0.875    0.875
    0.375    0.625    0.875
    0.375    0.875    0.625
    0.500    0.500    0.500
    0.500    0.750    0.750
    0.750    0.500    0.750
    0.750    0.750    0.500
    0.625    0.625    0.625
    0.625    0.875    0.875
    0.875    0.625    0.875
    0.875    0.875    0.625
&END

&DFT
  FUNCTIONAL LDA
&END  


More information about the CPMD-list mailing list