BLAS/LAPACK¶

Intel MKL¶

We recommend using the Intel MKL implementation of maths libraries. For this, you will need to load the intel-mkl module:

module load 2022r2
module load intel-mkl

And, depending on your application/compiler, export correct environment variables:

export PATH=/apps/arch/2022r2/software/linux-rhel8-skylake_avx512/gcc-8.5.0/intel-mkl-2020.4.304-562z3j76h4zy26bjp5r2mrimud6fshrc/mkl/bin:$PATH
export CPATH=/apps/arch/2022r2/software/linux-rhel8-skylake_avx512/gcc-8.5.0/intel-mkl-2020.4.304-562z3j76h4zy26bjp5r2mrimud6fshrc/mkl/include:$CPATH
export LIBRARY_PATH=/apps/arch/2022r2/software/linux-rhel8-skylake_avx512/gcc-8.5.0/intel-mkl-2020.4.304-562z3j76h4zy26bjp5r2mrimud6fshrc/mkl/lib/intel64:$LIBRARY_PATH
export LD_LIBRARY_PATH=/apps/arch/2022r2/software/linux-rhel8-skylake_avx512/gcc-8.5.0/intel-mkl-2020.4.304-562z3j76h4zy26bjp5r2mrimud6fshrc/mkl/lib/intel64:$LD_LIBRARY_PATH

After this, BLAS and LAPACK should be picked up from the Intel MKL implementation.

Note

The intel-mkl module only contains maths libraries. If you are looking for Intel compilers, you should load the full intel/oneapi-all suite of Intel goodness:

module load intel/oneapi-all

If you are looking for FFTW, you should check out this guide.