[FLASH-USERS] first model: hdf5 error

Seyit Hocuk seyit at mpe.mpg.de
Tue Oct 10 11:10:22 EDT 2017


Hi,

Try installing hdf5 without the enable-parallel. The serial should work 
just fine.
The parallel hdf5 option is probably more relevant for those who have 
very large simulations (> 10,000 CPU).

Best,
Seyit



On 10/10/17 16:56, Peter Woitke wrote:
>
> Ah sorry, I see, thank you Ryan and John.
>
> Unfortunately, this is not the solution. I already had
>
>
> F90FLAGS = -I ${HDF5_PATH}/include -DH5_USE_16_API
> CFLAGS_HDF5  = -I${HDF5_PATH}/include -DH5_USE_16_API
>
> in my Makefile.h inflash4.4/sites/prodimo. I have attached my complete 
> Makefile.h
>
>
>   Peter
>
> ------------------------------------------------------------------------
> *From:* Ryan Farber <rjfarber at umich.edu>
> *Sent:* 10 October 2017 15:50
> *To:* Peter Woitke
> *Cc:* Mark Richardson; flash-users
> *Subject:* Re: [FLASH-USERS] first model: hdf5 error
> Hi Peter,
>
> I believe Mark was referring to Makefile.h in the sites directory. 
> Unless you are on a system known by FLASH, the relevant Makefile.h 
> should be located at
>
> /PATH_TO_FLASH/sites/Prototypes/Linux/Makefile.h
>
> Best,
> --------
> Ryan
>
> On Tue, Oct 10, 2017 at 7:47 AM, Peter Woitke <pw31 at st-andrews.ac.uk 
> <mailto:pw31 at st-andrews.ac.uk>> wrote:
>
>     Hi Mark,
>
>
>     thanks for that. However, I cannot find any Makefile.h, not before
>     nor after ./configure --prefix=/home/star/pw31/software/hdf5
>     --enable-fortran --enable-parallel.
>
>     All I see is a Makefile.in and a Makefile, but none of them have
>     entries as you listed.
>
>     Any idea?
>
>
>       Peter
>
>
>     ------------------------------------------------------------------------
>     *From:* Mark Richardson <mark.richardson.work at gmail.com
>     <mailto:mark.richardson.work at gmail.com>>
>     *Sent:* 10 October 2017 15:38
>     *To:* Peter Woitke
>     *Cc:* flash-users
>     *Subject:* Re: [FLASH-USERS] first model: hdf5 error
>     Hi Peter,
>
>       it might be because you haven't flagged HDF5 to use the version
>     1.6 setup. In my Makefile.h I add DH5_USE_16_API to the following
>     definitions:
>          F90FLAGS
>          CFLAGS_HDF5
>     Perhaps that will solve your problem,
>        -Mark
>
>     On 10 October 2017 at 10:33, Peter Woitke <pw31 at st-andrews.ac.uk
>     <mailto:pw31 at st-andrews.ac.uk>> wrote:
>
>         Hi there,
>
>         I just downloaded and compiled mpich, hdf5 and flash, and I am
>         getting run-time error ...
>
>         HDF5-DIAG: Error detected in HDF5 (1.10.1) MPI-process 0:
>           #000: H5Ddeprec.c line 128 in H5Dcreate1(): not a location ID
>             major: Invalid arguments to routine
>             minor: Inappropriate type
>           #001: H5Gloc.c line 251 in H5G_loc(): invalid object ID
>             major: Invalid arguments to routine
>             minor: Bad value
>         flash4: io_h5create_dataset.c:57: io_h5create_dataset:
>         Assertion `dataset
>
>             = 0'
>
>
>         I have tested my mpich and hdf5 installations and they have
>         passed all tests. I am listing my installation commands below.
>         Maybe "enable-parallel" is not a good idea? My runSedov
>         directory has
>         ugo+rw permissions.
>
>         I read in another FLASH-USER entry that one should try
>
>             ./setup -auto Sedov +ug
>
>
>         This results in another run-time error ...
>
>         [Driver_init] Must set runtime parameters iProcs, jProcs,
>         kProcs, and meshCopyCount so that
>         iProcs*jProcs*kProcs*meshCopyCount equals number of processors
>         Calling MPI_Abort() for immediate shutdown!
>
>         Thank you for your help,
>
>           Peter Woitke
>
>
>         ----------------------------------------------------------
>
>             cd ~/software/hdf5-1.10.1
>             setenv CC  mpicc
>             setenv CXX mpicxx
>             setenv FC  mpif90
>             ./configure --help
>             ./configure --prefix=/home/star/pw31/software/hdf5
>
>                       --enable-fortran --enable-parallel
>
>             make
>             make check
>             make install
>             make check-install
>
>         add /home/star/pw31/software/hdf5/lib to LD_LIBRARY_PATH
>         ----------------------------------------------------------
>
>             cd ~/flash4.4/sites
>             mkdir prodimo
>             cp Prototypes/Linux/Makefile.h.in
>             <http://Makefile.h.in>tel prodimo/Makefile.h
>             cd prodimo
>             emacs Makefile.h ...
>             cd ~/flash4.4
>             ./setup -auto Sedov cd object
>             make
>
>         SUCCESS
>
>             mkdir ../runSedov
>             cd ../runSedov
>             cp
>             ../source/Simulation/SimulationMain/SedovChamber/flash.par .
>             mpirun -np 7 ../object/flash4
>
>         ----------------------------------------------------------
>
>
>
>         Dr. Peter Woitke
>         St Andrews University
>         School of Physics & Astronomy
>         top floor, room 306
>         North Haugh
>         KY16 9SS
>         St Andrews, UK
>         Tel: (+44) 1334 46 1681 <tel:%28%2B44%29%201334%2046%201681>
>         Fax: (+44) 1334 46 3104 <tel:%28%2B44%29%201334%2046%203104>
>
>
>
>
>     -- 
>
>     Mark Richardson
>     MAT Postdoctoral Fellow
>     Department of Astrophysics
>     American Museum of Natural History
>     Mark.Richardson.Work at gmail.com <mailto:Mark.Richardson.Work at gmail.com>
>     My Website <https://sites.google.com/site/marklarichardson/>
>     212 496 3432 <tel:%28212%29%20496-3432>
>
>

-- 
Dr. Seyit Hocuk
Post-doctoral researcher
Center for Astrochemical Studies (CAS)
Max-Planck-Institute for Extraterrestrial physics (MPE)
Giessenbachstrasse 1, 85748 Garching bei Munchen
Phone: +49-8930000-3012 | Fax: +49-8930000-3569
Web  : www.mpe.mpg.de/~seyit

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20171010/ba345db6/attachment.htm>


More information about the flash-users mailing list