oneAPI Math Kernel Library

The oneAPI Specification calls oneMKL a comprehensive math kernel library and solver

Definitions of SYCL/DPC++ interfaces for performance math library functions

Intel implements the specification as oneAPI Math Kernel Library

The Intel oneAPI Base Toolkit includes this free binary-only download. oneMKL is highly tuned for Intel GPUs and CPUs

The project demonstrates how any math library and target hardware can implement oneMKL SYCL/DPC++. It supports CPUs, GPUs, FPGAs, and accelerators

ONEAPI Math Kernel Library provides Linear Algebra, BLAS, LAPACK, SPARSE, FFT, RNG, Data Fitting, Vector Math, and Summary Statistics

GPGPU compute models connect hosts to GPUs. Each computer has several GPU computation Engines (CE), sometimes called Execution Units (EU) or Xe Vector Engines

oneAPI Math Kernel Library allows automated GPU offload dispatch with OpenMP pragmas, but we’ll focus on SYCL queues