Lightening-Transformer: A Dynamically-operated Optically-interconnected Photonic Transformer Accelerator

TL;DR

This paper introduces Lightening-Transformer, a novel photonic accelerator with dynamically-operated cores that significantly improves energy efficiency and latency for Transformer models, outperforming prior photonic and electronic accelerators.

Contribution

It presents the first dynamically-operated photonic tensor core and an integrated accelerator design tailored for Transformer workloads, enabling high performance and energy efficiency.

Findings

Achieves >2.6x energy reduction compared to prior photonic accelerators.

Achieves >12x latency reduction compared to prior photonic accelerators.

Delivers 2-3 orders of magnitude lower energy-delay product than electronic Transformer accelerators.

Abstract

The wide adoption and significant computing resource of attention-based transformers, e.g., Vision Transformers and large language models (LLM), have driven the demand for efficient hardware accelerators. There is a growing interest in exploring photonics as an alternative technology to digital electronics due to its high energy efficiency and ultra-fast processing speed. Photonic accelerators have shown promising results for CNNs, which mainly rely on weight-static linear operations. However, they encounter issues when efficiently supporting Transformer architectures, questioning the applicability of photonics to advanced ML tasks. The primary hurdle lies in their inefficiency in handling unique workloads in Transformers, i.e., dynamic and full-range tensor multiplication. In this work, we propose Lightening-Transformer, the first light-empowered, high-performance, and energy-efficient photonic Transformer accelerator. To overcome prior designs' fundamental limitations, we introduce a novel dynamically-operated photonic tensor core, DPTC, a crossbar array of interference-based optical vector dot-product engines supporting highly parallel, dynamic, and full-range matrix multiplication. Furthermore, we design a dedicated accelerator that integrates our novel photonic computing cores with photonic interconnects for inter-core data broadcast, fully unleashing the power of optics. Comprehensive evaluations show that ours achieves >2.6x energy and >12x latency reductions compared to prior photonic accelerators and delivers the lowest energy cost and 2 to 3 orders of magnitude lower energy-delay product compared to electronic Transformer accelerators, all while maintaining digital-comparable accuracy. Our work highlights the immense potential of photonics for advanced ML workloads, such as Transformer-backboned LLM. Our work is available at https://github.com/zhuhanqing/Lightening-Transformer.