This new core has been designed to enhance the performance of deep learning by 10X and enable more neural network processing per square millimetre.
