tests/QS/benchmark/H2O-32.inp failure in cp_fm_cholesky:cp_fm_cholesky_decompose

[Nick Wilson]

Hi Andy,

No, I haven't found a solution yet.

I confirmed it's coming out of dpotrf with info=30 which denotes the
Cholesky decomposition is failing.

I checked that the matrix passed didn't have any NaNs or things like
that but haven't done any analysis to see why the decomposition failed
or traced back to where the matrix is generated to see if I can spot
what's going wrong.

Nick


On Mar 16, 10:21 am, Andy Van Yperen-De Deyne
<andy.vanyp... at ugent.be> wrote:
> Hi Nick Wilson,
>
> I have the same problem when compiling with intel compiler and MKL
> libraries... Did you find a solution for this problem?
> When I compile with g95 everything seems to work (no
> cp_fm_cholesky_decompose problem) but it is MUCH slower compared to
> old versions compiled with intel.
>
> sincerely,
>
> Andy
>
> On Mar 2, 11:39 am, Nick Wilson <Nichola... at uk.fujitsu.com>
> wrote:
>
> > Hi,
>
> > I'm trying to run the tests/QS/benchmark/H2O-32.inp input file but
> > it's failing in cp_fm_cholesky:cp_fm_cholesky_decompose.
>
> > The Linux-x86-64-intel.sopt config file below is for a Xeon Woodcrest
> > with Intel compiler 11.1.069 and MKL but I've also tried various
> > combinations of serial/parallel, compiler versions and options, blas
> > and fft and they all fail with the same error.
>
> > An old version of the CP2K source downloaded on 7th July last year
> > works OK.
>
> > Any help working out the problem would be appreciated.
>
> > Thanks,
> > Nick Wilson
>
> > ==== Error message from  tests/QS/benchmark/H2O-32.inp ====
>
> >  ulimit -s unlimited
> >  ../../../exe/Linux-x86-64-intel/cp2k.sopt H2O-32.inp
> > ... snip ...
> >  Extrapolation method: initial_guess
> >  *
> >  *** 09:56:44 ERRORL2 in cp_fm_cholesky:cp_fm_cholesky_decompose
> > processor  ***
> >  ***      0  err=-300  condition FAILED at line
> > 92                          ***
> >  *
>
> >  ===== Routine Calling Stack =====
>
> >             9 cp_fm_cholesky_decompose
> >             8 make_basis_sm
> >             7 calculate_first_density_matrix
> >             6 scf_env_initial_rho_setup
> >             5 init_scf_run
> >             4 qs_energies_scf
> >             3 qs_forces
> >             2 qs_mol_dyn_low
> >             1 CP2K
> >  CP2K| condition FAILED at line 92
> >  CP2K| Abnormal program termination, stopped by process number 0
> > abort:
>
> > ==== arch/Linux-x86-64-intel.sopt ====
>
> > CC       = cc
> > CPP      =
>
> > INTEL_MKL = /share/apps/intel/fce/11.1.069/mkl
> > INTEL_INC = $(INTEL_MKL)/include/fftw
> > INTEL_LIB = $(INTEL_MKL)/lib/em64t
>
> > FC = ifort
> > LD = ifort
> > AR = xiar -r
>
> > DFLAGS = -D__INTEL -D__FFTSG -D__FFTW3 -D__FFTMKL
> > CPPFLAGS = -C -traditional $(DFLAGS) -I$(INTEL_INC)
> > FCFLAGS = $(DFLAGS) -I$(INTEL_INC) -O2 -xHost -heap-arrays 64 -fpp -
> > free
> > LDFLAGS = $(FCFLAGS)
> > LIBS = -L$(INTEL_LIB) -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -
> > lfftw3xf_intel
>
> > OBJECTS_ARCHITECTURE = machine_intel.o