Contains definitions for custom spack packages.
git clone https://github.com/nicspalla/my-repo.git <repo_name>
please replace the string <repo_name> here below and following with a name useful for you.
- spack version 0.16+
- GCC compilers suite version 9.0+
- installable with spack:
spack install [email protected] spack compiler add $(spack location -i [email protected])
- installable with spack:
spack config --scope user edit repos
then add the following two lines:
repos:
- /path/to/<repo_name>
please replace /path/to/<repo_name> with the complete path of this repository.
spack install yambo
This command will install the latest stable version of Yambo, with MPI enabled, using the spack preferred versions of the required libraries (openmpi, hdf5, netcdf-c, netcdf-fortran, fftw, libxc, lapack).
spack info yambo
This command displays the list of the supported versions of Yambo and the list of variants available.
Here a list of commands that will lead to a good installation of Yambo starting from the installation of Spack:
git clone https://github.com/nicspalla/my-repo.git $HOME/my-repo
git clone https://github.com/spack/spack.git $HOME/spack
cd $HOME/spack
git checkout releases/v0.18
. $HOME/spack/share/spack/setup-env.sh
spack external find
spack install [email protected] && spack compiler add $(spack location -i [email protected])
printf "repos:\n - $HOME/my-repo\n" > $HOME/.spack/repos.yaml
spack install yambo %[email protected] +mpi +openmp profile=time,memory
Now you can load the Yambo package and check if it works:
spack load yambo
yambo -h
Here an example of command installation of Yambo supporting acceleration with CUDA-Fortran on a workstation with NVidia Titan V devices (cc70):
spack install [email protected]+cuda-fortran+mpi+openmp+parallel_io+ph+rt+sc+nl cuda_arch=70 linalg=parallel,slepc %nvhpc
For MacOS users it is suggested to install Yambo with the GCC compiler, but using CMake compiled with the Clang compiler:
spack install [email protected]%[email protected] +mpi +openmp profile=memory,time ^cmake%apple-clang