\item {\bf ScaLAPACK}: ScaLAPACK library is used by DFT-FE via ELPA for its parallel linear algebra routines involving dense matrices, as well being a dependency for ELPA. \textcolor{red}{\bf If Intel MKL math library is available, please skip this step, as the ScaLAPACK libraries therein can be used directly.} If Intel MKL math library is not available, Netlib ScaLAPACK \url{http://www.netlib.org/scalapack/} needs to be installed using the instructions below. Download the current release version (2.2.0) from \url{http://www.netlib.org/scalapack/#\_software}, and build a shared library (use \verb|BUILD_SHARED_LIBS=ON|, \verb|BUILD_STATIC_LIBS=OFF| and \verb|BUILD_TESTING=OFF|) installation of ScaLAPACK using cmake. We recommend using the ccmake gui interface for the installation. Further, use the appropriate compilers for \verb|CMAKE_C_COMPILER| and \verb|CMAKE_FORTRAN_COMPILER|, and also use \verb|-fPIC| flag for \verb|CMAKE_C_FLAGS| and \verb|-fPIC -fallow-argument-mismatch| for \verb|CMAKE_Fortran_FLAGS|. For best performance, ScaLAPACK must be linked to optimized BLAS-LAPACK libraries by using\\ \verb|USE_OPTIMIZED_LAPACK_BLAS=ON|, and providing external paths to BLAS-LAPACK libraries (MKL, OpenBlas, ESSL etc.) during the cmake configuration.
0 commit comments