P-ReTiNA: Photonic Tensor Core Based Real- Time AI
Dharanidhar Dang · 2023
Computing deep AI algorithms on traditional CPUs and GPUs brings several performance and energy pitfalls. Several novel approaches based on ASIC, FPGA, and resistive-memory devices have been recently demonstrated with promising results. Most of them target only the inference (testing) phase of deep learning. There have been very limited attempts to design a full-fledged AI accelerator capable of both training and inference in realtime. It is due to the highly compute and memory intensive nature of the training phase. In this paper, we propose P- ReTiNA, a novel analog photonics CNN accelerator. P- ReTinA uses silicon microdisk-based convolution, photonic phase change memory-based memory, and dense- wavelength-division-multiplexing for energy-efficient and ultrafast deep learning. We evaluate P-ReTinA using a commercial CAD framework (IPKISS) on deep learning benchmark models including LeNet and VGG-Net. Compared to the state-of-the-art, P-ReTinA improves the CNN throughput, energy-efficiency, and computational efficiency by up to 48x, 45x, and 12x respectively with trivial accuracy degradation.