Abstract: The rapid advancement of deep learning, with increasingly large deep neural networks (DNNs), has led to the use of multi-core parallel processing in accelerators, utilizing Network-on-Chip ...