[FLASH-BUGS] problems running flash3

[Christopher Daley]

Hi Manami, 

It looks like FLASH is failing on the first floating point MPI
communication, so I suspect that you have not promoted Fortran
reals to double precision.  If you are using gfortran compiler
then the FFLAGS macros in your FLASH Makefile.h should have
-fdefault-real-8 -fdefault-double-8 options.  Do you have these
compiler options in your Makefile.h?

Chris


On Mon, 2011-01-24 at 08:06 +0100, Manami Sasaki wrote:
> Hi,
> 
> I have downloaded FLASH3.3 and also installed mpich and hdf5.
> > ./setup Sedov -auto
> works fine. However, if I try to run flash with
> > mpirun -np 1 /home/sasaki/soft/flash/FLASH3.3/object/flash3
> I get the result shown below and could not figure out how to
> revolve this. I'm working on a PC with openSUSE 11.1 (Linux
> 2.6.27.56-0.1-default x86_64) installed.
> 
> Thank you for your help.
> Best regards,
> Manami
> 
> 
> ---------------------------------------------------------------
> Fatal error in MPI_Allreduce: Internal MPI error!, error stack:
> MPI_Allreduce(826).......: MPI_Allreduce(sbuf=0x7fffa0b3d938,
> rbuf=0x7fffa0b3d93c, count=1, MPI_DOUBLE_PRECISION, MPI_MIN, 
> MPI_COMM_WORLD)
> failed
> MPIR_Allreduce_impl(684).:
> MPIR_Allreduce_intra(279):
> MPIR_Localcopy(349)......: memcpy arguments alias each other,
> dst=0x7fffa0b3d93c src=0x7fffa0b3d938 len=8
>   WARNING : Adaptive Grid did not find any refinement variables
>   MaterialProperties initialized
>   Cosmology initialized
>   Source terms initialized
>    iteration, no. not moved =            0           0
>   refined: total leaf blocks =            1
>   refined: total blocks =            1
>    starting MORTON ORDERING
>    tot_blocks after            1
>    max_blocks 2           1
>    min_blocks 2           1
>   INFO: Grid_fillGuardCells is ignoring masking.
>    Finished with Grid_initDomain, no restart
>   Ready to call Hydro_init
>   Info: Hydro_init has set hy_specialFluxVar to            3
>   Hydro initialized
>   Gravity initialized
> rank 0 in job 9  aitp28_40893   caused collective abort of all ranks
>    exit status of rank 0: killed by signal 9