Two oneMKL implementations

The Intel(R) oneAPI Math Kernel Library (oneMKL) Product as part of OneApi

only support Intel CPU && GPUs

icpx as CXX_COMPILER && icx as C_COMPILER

The oneAPI Math Kernel Library (oneMKL) Interfaces Project on github

give backend support for AMD && Nvidia GPUs

clang++ in intel/llvm for cuda backend support

Interfaces defined different from the first one

blas,lapack,rng

Documentation has errors

install

as told in uconfig

use

clang++ for cuda backend
add compile options:"-fsycl -fsycl-targets=nvptx64-nvidia-cuda -Xsycl-target-backend --cuda-gpu-arch=sm_86"
link libraties:"libonemkl.so" (will include libonemkl_blas_culas.so ... etc.)

sample

a oneMKL,oneDNN sample