CrossLight: A Cross-Layer Optimized Silicon Photonic Neural Network Accelerator

TL;DR

CrossLight is a novel silicon photonic neural network accelerator that significantly improves energy efficiency and throughput through cross-layer optimization and device engineering, outperforming existing photonic accelerators.

Contribution

It introduces a comprehensive cross-layer design approach for silicon photonic neural network accelerators, combining device, circuit, and architecture-level innovations.

Findings

9.5x lower energy-per-bit compared to state-of-the-art

15.9x higher performance-per-watt at 16-bit resolution

Enhanced resilience to process variations and thermal crosstalk

Abstract

Domain-specific neural network accelerators have seen growing interest in recent years due to their improved energy efficiency and inference performance compared to CPUs and GPUs. In this paper, we propose a novel cross-layer optimized neural network accelerator called CrossLight that leverages silicon photonics. CrossLight includes device-level engineering for resilience to process variations and thermal crosstalk, circuit-level tuning enhancements for inference latency reduction, and architecture-level optimization to enable higher resolution, better energy-efficiency, and improved throughput. On average, CrossLight offers 9.5x lower energy-per-bit and 15.9x higher performance-per-watt at 16-bit resolution than state-of-the-art photonic deep learning accelerators.