[Octopus-users] Segmentation fault with SpinComponents = spin_polarized

[Myrta Gruning]

Thank you Abram.

Yes, it helped either lowering to -O0 for both ifort and icc or lowering
ifort to -O0 and using gcc. But I think this is a bit too drastic... ;-)

Cheers,

m	 

On Wed, 2008-12-17 at 09:33 -0700, Abram Van Der Geest wrote:
> Hello,
> 
> I have seen the same thing when using the Intel compiler.  I got  
> around it by using the debug compiler flag.
> 
> Hope this helps.
> 
> Bram
> 
> 
> On Dec 17, 2008, at 7:18 AM, Myrta Gruning wrote:
> 
> > Hallo,
> >
> > I have got a Segmentation Fault when using SpinComponents =
> > spin_polarized in my job (I have tested different systems and  
> > parameters
> > always getting the same error).
> > It is strange since it appears just on a particular machine (on other
> > machines the same test jobs went without problems) and is probably due
> > to some library, compiler or compiler options (or combination).
> >
> > Thanks,
> >
> > m
> >
> > ----------------
> > Here follow the details:
> >
> > *inp (just a a test, same behavior for other systems I have tried):
> >
> > CalculationMode = gs
> > Units = eV_Angstrom
> > FromScratch = yes
> >
> > spacing = 0.15
> > Radius = 4
> >
> > %Coordinates
> >  'C' | 0.0 | 0.0 | 0.0 |
> > %
> >
> > ConvAbsDens = 1e-7
> > SpinComponents = spin_polarized
> >
> > *error message
> > [...]
> > Input: [RelativisticCorrection = non_relativistic]
> > Input: [TDGauge = length]
> > Input: [AbsorbingBoundaries = no_absorbing]
> > Input: [GuessMagnetDensity = ferromagnetic]
> > Info: Unnormalized total charge =      5.999923
> > Info: Renormalized total charge =      6.000000
> > Info: Setting up Hamiltonian.
> > [green:04588] *** Process received signal ***
> > [green:04588] Signal: Segmentation fault (11)
> > [green:04588] Signal code: Address not mapped (1)
> > [green:04588] Failing at address: 0x38
> > [green:04588] [ 0] /lib64/libpthread.so.0 [0x3a8100de80]
> > [green:04588]
> > [ 1] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(xc_m_mp_xc_get_vxc_+0xc6c) [0x5ea1f6]
> > [green:04588] [ 2] /home/pcpm/gruning/octopus-3.0.1/bin/octopus_mpi
> > [0x5ac11e]
> > [green:04588]
> > [ 3] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(v_ks_m_mp_v_ks_calc_+0x585) [0x5ab551]
> > [green:04588]
> > [ 4] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(ground_state_m_mp_ground_state_run_+0x215) [0x524ebd]
> > [green:04588]
> > [ 5] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(run_prog_m_mp_run_
> > +0x308) [0x45ff1a]
> > [green:04588]
> > [ 6] /home/pcpm/gruning/octopus-3.0.1/bin/octopus_mpi(MAIN__+0x901)
> > [0x460925]
> > [green:04588] [ 7] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(main
> > +0x2a) [0x444412]
> > [green:04588] [ 8] /lib64/libc.so.6(__libc_start_main+0xf4)
> > [0x3a8081d8b4]
> > [green:04588]
> > [ 9] /home/pcpm/gruning/octopus-3.0.1/bin/ 
> > octopus_mpi(gsl_complex_arcsec
> > +0x51) [0x444339]
> > [green:04588] *** End of error message ***
> > Segmentation fault
> >
> > * machine+OS
> > [gruning at green Test]$ uname -a
> > Linux green 2.6.18-92.1.18.el5 #1 SMP Wed Nov 12 06:45:11 EST 2008
> > x86_64 x86_64 x86_64 GNU/Linux
> >
> > * compiler for octopus: ifort/icc 10.1
> >                        (mpi enabled) openmpi
> > * note: all libs, but gsl were on the system already, compiled with
> > intel compiler. I compiled the gsl 1.10. I was forced to use gcc  
> > (4.1.2)
> > since with icc I did not pass the checks.
> >
> >
> >
> > _______________________________________________
> > Octopus-users mailing list
> > Octopus-users at tddft.org
> > http://www.tddft.org/mailman/listinfo/octopus-users
>