Overview of the New Intel® oneAPI Math Kernel Library (oneMKL)
The Intel Math Kernel Library has long provided a wide range of mathematical functionality optimized for Intel CPUs. With Intel’s upcoming launch of discrete GPUs and the Aurora system at Argonne National Laboratory, the introduction of the oneMKL product adds support for Intel GPUs to the existing CPU-optimized implementations. oneMKL is part of Intel’s oneAPI, an environment for heterogeneous computing based around the DPC++ programming language.
The presentation will begin with a brief overview of the functionality available in oneMKL, the different programming models it supports for CPU- and GPU-based computation (DPC++ and OpenMP offload), and how to use its asynchronous and batch capabilities. There will be plenty of time for audience participation and questions about oneMKL.
Peter Caday is a Math Algorithms Engineer at Intel, working primarily on BLAS optimization. Prior to joining Intel, he was a postdoctoral fellow at Rice University’s Computational and Applied Mathematics Department, where his research focused on microlocal analysis and geophysical imaging.