With the increasing demand for computing power of AI applications, CPU, GPU and dedicated acceleration, chips can all work together through heterogeneous computing to enhance the computing power of the system. We collectively call them XPU.
Intel continues to add new features to enhance AI computing in its newly launched CPUs and GPUs, such as Intel® Deep Learning Boost, Intel® Advanced Matrix Extensions, and Intel® Xe Matrix Extensions technologies. These new features can greatly accelerate the performance of AI training and inference.
In order to facilitate users to use these hardware features, Intel also launched the Intel® AI Analytics Toolkit software toolkit, which includes AI frameworks and middleware optimized for new hardware features, and is free for users to use. Among them, there are AI framework performance enhancement tools provided in the form of plug-ins such as Intel Extension® for Tensorflow*/Pytorch*/Scikit-learn*.