A Silicon Photonic Accelerator for Convolutional Neural Networks with Heterogeneous Quantization
This work addresses the need for efficient and accurate CNN inference for applications in edge computing or data centers, representing an incremental improvement in hardware acceleration.
The paper tackles the problem of accuracy degradation in homogeneously quantized CNNs by proposing HQNNA, a silicon photonic accelerator that supports heterogeneous quantization, achieving up to 73.8x better energy-per-bit and 159.5x better throughput-energy efficiency compared to state-of-the-art photonic accelerators.
Parameter quantization in convolutional neural networks (CNNs) can help generate efficient models with lower memory footprint and computational complexity. But, homogeneous quantization can result in significant degradation of CNN model accuracy. In contrast, heterogeneous quantization represents a promising approach to realize compact, quantized models with higher inference accuracies. In this paper, we propose HQNNA, a CNN accelerator based on non-coherent silicon photonics that can accelerate both homogeneously quantized and heterogeneously quantized CNN models. Our analyses show that HQNNA achieves up to 73.8x better energy-per-bit and 159.5x better throughput-energy efficiency than state-of-the-art photonic CNN accelerators.