As computing demand and memory footprint of deep learning applications accelerate,
clusters of cores sharing local (L1) multi-banked memory are widely used as key building …