Problem with thorn EOS_Omni

Comments (6)

1. 

   Roland Haas

   • removed comment

   Hello Maxim,

   EOS_Omni requires Fortran bindings to libhdf5 (libhdf5_fortran.so and .a) which some Linux distributions do no install by default. You also need to activate the thorn

   ExternalLibraries/HDF5 in your thorn list. You must also make sure that hdf5.mod is in '''include''' path of the Fortran compiler (http://mpc.uci.edu/man/gfortran.html). If HDF5 is already active, try adding

   HDF5_DIR=BUILD to you config file (/home/bmv/.cactus/config it seems). Lastly you should make sure that you use the same compiler (and version) to build Cactus that was used to compile libhdf5 (not an issue if you use BUILD).

   Yours, Roland

   • 2011-10-04T10:19:18+00:00
2. 

   anonymous reporter

   • removed comment

   Thank you Roland,

   I have modify /home/bmv/.cactus/config by adding HDF5 = yes HDF5_DIR=BUILD

   HDF5 is in the throns list. .... EinsteinUtils/TGRtensor # TGRtensor ( ) [ ] { } ExternalLibraries/BLAS # BLAS ( ) [ ] { } ExternalLibraries/GSL # GSL ( ) [ ] { } ExternalLibraries/HDF5 # HDF5 ( ) [ ] { } ExternalLibraries/LAPACK # LAPACK ( ) [ ] { } ExternalLibraries/LORENE # LORENE ( ) [ ] { } ExternalLibraries/OpenSSL # OpenSSL ( ) [ ] { } ExternalLibraries/libjpeg # libjpeg ( ) [ ] { } ExternalLibraries/zlib # zlib ( ) [ ] { } ....

   Unfortunately, the mistake still on it's place.

   Preprocessing /home/bmv/utils/soft/develop/ET/Cactus/arrangements/EinsteinEOS/EOS_Omni/src/nuc_eos/readtable.F90 Compiling /home/bmv/utils/soft/develop/ET/Cactus/arrangements/EinsteinEOS/EOS_Omni/src/nuc_eos/readtable.F90 /home/bmv/utils/soft/develop/ET/Cactus/configs/test1/build/EOS_Omni/nuc_eos/readtable.f90:5.10:

   use hdf5 1 Fatal Error: Can't open module file 'hdf5.mod' for reading at (1): No such file or directory make[3]: * [readtable.F90.o] Error 1 make[2]: * [nuc_eos/make.checked] Error 2 make[1]: * [/home/bmv/utils/soft/develop/ET/Cactus/configs/test1/lib/libthorn_EOS_Omni.a] Error 2 make: * [test1] Error 2

   Is it possible to show for the compiler the path to hdf5.mod in such configuration? (hdf5.mod not exist in the Cactus folder)

   Best wishes, Maxim

   • 2011-10-04T10:58:21+00:00
3. 

   Roland Haas

   • removed comment

   Hello Maxim,

   HDF5=yes should no longer be used (in fact it should generate a warning). Though this should not give you the errors you see.

   I would suggest doing a `make test1-realclean` to make sure everything is rebuild (the ExternalLibraries are a bit odd that way). You should then see some messages from HDF5 during the initial compilation phase (CST configure I believe is the technical term), which should end in something like `HDF5_LIBRARIES = hdf5 hdf5_fortran hdf5_cxx ...`. Otherwise I have had success by manually (really bad idea usually, but it worked) copying hdf5.mod into Cactus/configs/test1/scratch .

   If it still fails after a realclean, would you mind attaching all output ie:

   make test1-realclean make test1-config options=/.cactus/config &>test1.log make test1 >>test1.log 2>>test1.log

   • 2011-10-04T11:18:31+00:00
4. 

   anonymous reporter

   • removed comment

   Thank you Roland,

   I do make test1-realclean and remove HDF5=yes After that ET was compiled successfully.

   Now I try to run mpirun np 16 exe/cactus_test1 par/static_tov.par

   Best wishes, Maxim

   • 2011-10-05T07:00:17+00:00
5. 

   Frank Löffler

   • changed status to resolved
   • removed comment

   • 2011-10-05T09:27:34+00:00
6. 

   Roland Haas

   • edited description
   • changed status to closed

   • 2019-02-21T20:23:44+00:00
7. Log in to comment