Technology Category
Machine Learning: Hardware-aware MLComputer Vision: Multi-modal VisionCognitive Modeling & Cognitive Systems: Neural Spike Coding
A 262 TOPS Hyperdimensional Photonic AI Accelerator powered by a Si3N4 microcomb laser
🤖 AI Summary
To address the energy-efficiency and computational throughput bottlenecks arising from digital AI accelerators approaching physical limits, this work proposes a multidimensional photonic neural network (PNN) architecture. It introduces, for the first time, time–wavelength–space-division multiplexing (T-WSDM) encoding, implemented on a Si₃N₄ microcomb laser integrated with a 16×16 arrayed-waveguide grating router (AWGR) photonic chip. The resulting high-parallel photonic AI accelerator achieves 262 TOPS of optical computational throughput—24× higher than state-of-the-art waveguide-based accelerators—and supports in-flight photonic tensor multiply-accumulate operations and on-chip multi-wavelength interferometric detection. Evaluated on DDoS attack detection, it attains a Cohen’s kappa of 0.867; on MNIST classification, it achieves 92.14% accuracy—both nearing software-level performance—while operating at a measured symbol rate of 32 Gbaud. The core innovations are the T-WSDM three-dimensional coordinated encoding scheme and the integrated microcomb-driven multidimensional photonic computing paradigm.
Technology Category
Application Category
📝 Abstract
The ever-increasing volume of data has necessitated a new computing paradigm, embodied through Artificial Intelligence (AI) and Large Language Models (LLMs). Digital electronic AI computing systems, however, are gradually reaching their physical plateaus, stimulating extensive research towards next-generation AI accelerators. Photonic Neural Networks (PNNs), with their unique ability to capitalize on the interplay of multiple physical dimensions including time, wavelength, and space, have been brought forward with a credible promise for boosting computational power and energy efficiency in AI processors. In this article, we experimentally demonstrate a novel multidimensional arrayed waveguide grating router (AWGR)-based photonic AI accelerator that can execute tensor multiplications at a record-high total computational power of 262 TOPS, offering a ~24x improvement over the existing waveguide-based optical accelerators. It consists of a 16x16 AWGR that exploits the time-, wavelength- and space- division multiplexing (T-WSDM) for weight and input encoding together with an integrated Si3N4-based frequency comb for multi-wavelength generation. The photonic AI accelerator has been experimentally validated in both Fully-Connected (FC) and Convolutional NN (NNs) models, with the FC and CNN being trained for DDoS attack identification and MNIST classification, respectively. The experimental inference at 32 Gbaud achieved a Cohen's kappa score of 0.867 for DDoS detection and an accuracy of 92.14% for MNIST classification, respectively, closely matching the software performance.
Problem
Research questions and friction points this paper is trying to address.
Develop photonic AI accelerator for high computational power
Enhance energy efficiency in AI processors using photonics
Validate accelerator in DDoS detection and MNIST classification
Innovation
Methods, ideas, or system contributions that make the work stand out.
262 TOPS photonic AI accelerator using Si3N4 microcomb
Multidimensional AWGR with T-WSDM for tensor operations
Validated in FC and CNN models for AI tasks
🔎 Similar Papers
No similar papers found.
