[CP2K:2193] ERRORL2 in qs_loc_methods:centers_spreads_berry processor

[Teodoro Laino]

Dear Berosek,

this error could pop up when some memory structures were not allocated 
(a general memory allocation problem).
Unfortunately the memory involved with that regtest is quite small.
Therefore I believe it is entirely a compiler bug. It is frequent that 
Intel fortran
miscompiles CP2K.
In fact, Version 10.0.025 seem to work with that regtest.

3 possible solutions:
-) use a lower optimization level rather than -O3 (most of the compiler 
bugs lie in a too aggressive/buggy optimization)
-) Try another version of the same compiler
-) Try a different compiler

Regards,
Teo

berosek wrote:
> Dear all,
> I got this message "ERRORL2 in qs_loc_methods:centers_spreads_berry"
> when I was running do_reg_test. It repeat approximately 103 times. For
> error reconstruction it first occurred during running test: QS/regtest-
> gapw-4/CO_xastpxfh.inp. Output can be seen here:
> http://groups.google.com/group/cp2k/web/CO_xastpxfh.inp.out and
> error_summary: http://groups.google.com/group/cp2k/web/error_summary .
>
> I have no idea how to solve this problem.
> I'm using intel compiler version 11.0.081 with mkl, fftw 3.2.1, libint
> 1.1.4 and openmpi 1.3.3. This is arch file I'm using:
>
> FFTW    = /home/install/cp2k/fftw-3.2.1/
> done
> FFTW_INC = $(FFTW)/
> include
> FFTW_LIB = $(FFTW)/
> lib
> MKLPATH = /opt/intel/Compiler/11.0/081/mkl/lib/
> em64t
> INTPATH = /opt/intel/Compiler/11.0/081/lib/
> em64t
> LIBINT  = /home/install/cp2k/libint-1.1.4/done/
> lib
> LIBINT_WRAPPER = /home/install/cp2k/test/cp2k/libint_tools/
> libint_cpp_wrapper.o
> MPI = /home/install/
> openmpi-1.3.3
> MPI_INC = $(MPI)/
> include
> MPI_LIB = $(MPI)/
> lib
> CC       = $(MPI)/bin/mpicc
> FC       = $(MPI)/bin/mpif77
> LD       = $(MPI)/bin/mpif77
> AR       = ar -r
> DFLAGS   = -D__INTEL -D__FFTSG -D__parallel -D__BLACS -D__SCALAPACK -
> D__LIBINT -D__FFTW3
> CPPFLAGS =
> FCFLAGS  = $(DFLAGS) -I$(FFTW_INC) -I$(INCLUDE) -I$(MPI_INC) -O3 -heap-
> arrays 64 -unroll -fpp -free -pc64 -march=pentium3 -mtune=core2
> FCFLAGS2  = $(DFLAGS) -I$(FFTW_INC) -I$(INCLUDE) -I$(MPI_INC) -O1 -
> heap-arrays 64 -fpp -free -unroll -pc64 -march=pentium3 -mtune=core2
> LDFLAGS  = $(FCFLAGS) -L$(MPI_LIB) -static-intel
> LIBS     = -L$(LIBRARY_PATH) -L$(MKLPATH) -L$(INTPATH) \
>            -lmkl_scalapack_lp64                        \
>            -lmkl_blacs_openmpi_lp64                    \
>            -lmkl_intel_lp64                            \
>            -lmkl_sequential                            \
>            -lmkl_core                                  \
>            $(LIBINT_WRAPPER)                          \
>            $(LIBINT)/libderiv.a                       \
>            $(LIBINT)/libint.a                         \
>            -lstdc++                                    \
>            $(FFTW_LIB)/libfftw3.a                      \
>            $(MPI_LIB)/libmpi.a                          \
>            $(MPI_LIB)/libmpi_f77.a
> OBJECTS_ARCHITECTURE = machine_intel.o
> graphcon.o: graphcon.F
>         $(FC) -c $(FCFLAGS2) $<
>
> I would be glad for any suggestions.
> Thanks Berosek
> >
>