Half-Precision Math in Modeling and Code Generation

Learn about the half-precision datatype in MATLAB®. Walk through the process of building highly efficient embedded algorithms and generating efficient C++, HDL, and GPU code that use half-precision. See how using half-precision can speed up your deep learning, computer vision, and other workflows. The half-precision datatype is best suited for applications that benefit from reduced memory usage and increased computation speed but do not require as much precision as double or single datatypes. It uses less memory than double or single datatypes while handling wider dynamic ranges than integer or fixed-point datatypes of the same size. Half-Precision Sobel Edge Detection Example: https://www.mathworks.com/help/releas... Half-Precision Field-Oriented Control Example: https://www.mathworks.com/help/releas...