A Multi-precision Multiply-Accumulation Array
Chaolin Rao, Yueyang Zheng, Haochuan Wan · 2022 19th International SoC Design Conference (ISOCC) · 2022
This paper presents a precision scalable multiply-accumulate (MAC) array for the implementation of Neural Radiance Field (NeRF) rendering for energy and circuit area reduction. The proposed MAC operates multiple precisions in different cycles. Since the operations on MAC are fixed, there is no other configurable logic, making it more energy efficient.