Dear Soumyajyotih,

Actually our group-internal cluster has that kind of CPUs. One has to be careful to compile Fleur in such a way that it shows good performance on such systems. Especially when you use Intel compilers and libraries this is not the use case for which they were developed. You can get inspiration for the compilation on such a system by having a look at the cmak/IFFAMD.sh machine file in your fleur directory:

echo "Using special config for IFF-cluster"
 
#Set the compilers to mpiifort, mpiicc, mpiicpc
export FC=mpiifort
export CC=mpiicc
export CXX=mpiicpc
 
FLEUR_LIBRARIES="-lxml2;-lmkl_scalapack_lp64;-lmkl_blacs_intelmpi_lp64;-mt_mpi"
#echo "Wannier is switched off-manually"
#CLI_USE_WANNIER=FALSE
echo "ChASE is switched off-manually"
CLI_USE_CHASE=FALSE
echo "Broken linker can be used"
CLI_WARN_ONLY=1
echo "Patch MKL"
CLI_PATCH_INTEL=1
 
#Set environment variables usefull for external dependencies, e.g. ELPA
export CFLAGS=-mkl
export CMAKE_Fortran_FLAGS="$CMAKE_Fortran_FLAGS -mkl"
export FCFLAGS=-mkl
export LIBS="-mkl -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64"
export SCALAPACK_LDFLAGS="-lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64"
 
 

Especially the variables defined in this machine file are important (e.g. CLI_PATCH_INTEL). They are used to execute certain code paths in the configuration script. I additionally use the compiler option "-amd" when calling the configure script, though I am unsure whether that is actually needed. It looks like this is just another way of setting CLI_PATCH_INTEL to 1.

You have to be aware that even though these AMD CPUs are compatible to Intel CPUs, their performance characteristics and memory layout differ. You can get more performance out of the AMD CPUs, but they are not really more performant per CPU core, but due to the availability of more cores. This means that you also have to make sure that you invoke Fleur with adequate parallelization schemes with a reasonable number of MPI processes and a reasonable number of OpenMP threads.