Intel oneDNN 2.0 Deep Neural Network Library Working On More Performance Tuning

… INT8 convolutions with 1×1 kernel and spatial strides, and a variety of other specific areas within this deep learning library seeing optimizations.

Link to Full Article: Read Here