automake to be installed.
It is recommended to install FLAME in a directory different from the source.
You first need to install
futilewhich is a set of utilities from the BigDFT suite. It is highly recommended to use the tar file of futile given inside the FLAME to avoid possible inconsistency.
You do NOT need different installation of futile for multiple installations of FLAME.
Here are steps:
Untar the futile inside the FLAME, then in order to build futile, make a directory somewhere and from there run the following:
path_to_futile_source/Installer.py build futile -c "FCFLAGS=-O2" "--with-ext-linalg=-L/opt/intel/composer_xe_2013.2.146/mkl/lib/intel64 -lmkl_rt -lmkl_scalapack_lp64 -lmkl_blacs_openmpi_lp64 -liomp5 -lm" "CC=mpicc" "CXX=mpicxx" "FC=mpif90" "F77=mpif90" "FCLIBS= "
- Of course, the setting is given for a particular version and installation of
MKL, you must need to modify it according to your machine.
- After installation, you may need to find out how to link with the build futile, however, in the subsequent steps you see an example. In case, you want to know how to link you can run:
path_to_futile_source/Installer.py link futile
- in FLAME source directory run:
- make a directory to build flame and then replace
$FUTILEby the full path of futile built directory or define it as an environment variable:
- IMPORTANT: exporting FUTILE variable is required, not optional.
path_to_flame_source/configure FC=mpif90 F77=mpif90 CXX=mpicc CC=mpicc FCFLAGS="-I$FUTILE/install/include -shared-intel -mcmodel=large -mkl=sequential" CFLAGS=-mcmodel=large "LIBS=-L$FUTILE/install/lib -L/opt/intel/composer_xe_2013.2.146/mkl/lib/intel64 -lfutile-1 -lmkl_rt -lmkl_scalapack_lp64 -lmkl_blacs_openmpi_lp64 -liomp5 -lm -lyaml -ldl -lrt -cxxlib"
- Linking to exterior packages:
- if you want to link with SPGLIB then add
- if you want to link with BigDFT PSolver then add
- if you want to link with LAMMPS then add
- You should be ready to use FLAME, the executable is at