Intel® Deep Link and Workload Throughput

Unlocking the Power of Intel® Deep Link Part Two: Increasing Workload Throughput. This paper is part two in a series of white papers designed to provide details regarding openly available development tools that can be used to take full advantage of Intel® Deep Link. This second paper focuses on applying Intel Deep Link concepts for improving workload throughput using various backend options, presents Leela Chess Zero as a use case, and introduces Intel® oneAPI Deep Neural Network Library (oneDNN): a high performance library for deep learning network applications.