Photonic Tensor Core

Photonic tensor cores (PTCs) are optical computing components designed to accelerate artificial intelligence (AI) computations, primarily matrix multiplications crucial for deep neural networks. Current research focuses on optimizing PTC design through automated topology search algorithms, algorithm-hardware co-design for sparsity exploitation, and time-multiplexing techniques to improve energy efficiency and reduce hardware footprint. These advancements aim to overcome limitations in reconfigurability, power consumption, and precision, ultimately enabling more efficient and powerful AI accelerators for both edge and cloud computing applications.