Bin,

Check the list archives for emails of the last week. Someone had a
problem similar to yours for a small system, it couls be something
related.

Marcos

On Thu, Jul 8, 2010 at 3:51 PM, Bin Shao <[email protected]> wrote:
> Dear siesta users,
>
> When I tried to do the Ex1 of Barcelona 07 - CH4, there came something wrong
> with setting the PAO.BasisSize to be DZP and running the siesta in parallel.
> And I compiled the siesta with impi. Here is the wrong message:
>
> stepf: Fermi-Dirac step function
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> Image              PC                Routine            Line        Source
> siesta             00000000004AA7ED  Unknown               Unknown  Unknown
> siesta             0000000000498890  Unknown               Unknown  Unknown
> siesta             00000000005E0FA2  Unknown               Unknown  Unknown
> siesta             000000000046316C  Unknown               Unknown  Unknown
> libc.so.6          0000003F9681D994  Unknown               Unknown  Unknown
> siesta             0000000000463079  Unknown               Unknown  Unknown
> rank 7 in job 1  node18c.emlab_58968   caused collective abort of all ranks
>   exit status of rank 7: return code 174
>
> and the input file:
>
> #General system specifications
> SystemName          CH4 molecule
> SystemLabel         ch4
> NumberOfAtoms       5
> NumberOfSpecies     2
>
> %block ChemicalSpeciesLabel
>  1  6 C   # Species index, atomic number, species label
>  2  1 H   # Species index, atomic number, species label
> %endblock ChemicalSpeciesLabel
>
>
> #Atomic coordinates
> AtomicCoordinatesFormat  Ang
>
> %block AtomicCoordinatesAndAtomicSpecies
>   0.000      0.000      0.000   1
>   1.219     -0.284     -0.377   2
>  -0.284      1.219     -0.377   2
>  -0.140     -0.140      1.219   2
>  -0.833     -0.833     -0.503   2
> %endblock AtomicCoordinatesAndAtomicSpecies
>
>
> # Basis set definition
> PAO.EnergyShift 250 meV
> PAO.SplitNorm   0.15
> PAO.BasisSize   DZP
>
> #Real space grid
> MeshCutoff 10.0 Ry
>
> # Convergence of SCF
> MaxSCFIterations 50
> DM.MixingWeight 0.9
> DM.NumberPulay  2
>
> # Type of solution (diagon is the  default for less than 100 atoms)
> SolutionMethod diagon
>
> If I set the PAO.BasisSize to be DZ or SZ or do not run the siesta in
> parallel, the calculation will finish successfully. In addition, there is 8
> nodes of my cluster and I run the program with the command mpirun -r ssh -np
> X siesta < ch4.fdf, (X is the number of the nodes I needed). If X < 8, the
> calculation will finish successfully even setting the PAO.BasisSize to be
> DZP.
>
> What's the problem? How can I run the program with 8 nodes, meanwhile
> setting the PAO.BasisSize to be DZP?
>
> Any suggestion will be appriciated. Thank you in advance!
>
> Best regards,
>
>
> --
> Bin Shao
> College of Information Technical Science, Nankai University
> 94 Weijin Rd. Nankai Dist. Tianjin 300071, China
> Email: [email protected]
>

Responder a