LightMat-HP: A Photonic-Electronic System for Accelerating General Matrix Multiplication With Configurable Precision

2026-04-14

AI Summary

This work addresses the limitations of conventional electronic accelerators—constrained by memory bandwidth and energy efficiency—and the inability of existing photonic computing systems to support high-precision general-purpose matrix multiplication due to analog noise and insufficient modulation accuracy. To overcome these challenges, the authors propose LightMat-HP, a hybrid electro-optical acceleration architecture that innovatively integrates block floating-point (BFP) representation, sliced photonic multiplication, and digital accumulation to enable configurable-precision, arbitrary-size matrix operations. Leveraging a tile-based dataflow and co-designed electro-optical mechanisms, LightMat-HP demonstrates superior performance over FPGAs, GPUs, and state-of-the-art photonic accelerators in both prototype validation and large-scale simulations, achieving exceptional throughput, low latency, and high energy efficiency—particularly for small-to-medium-scale matrix multiplications.

Technology Category

• Machine Learning: Hardware-aware ML
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Non-convex Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web applications in cross-disciplinary domains and verticals such as mixed reality, smart cities, and digital health
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Matrix multiplication is a fundamental kernel in large-scale artificial intelligence and scientific computing, but its performance on conventional electronic accelerators is increasingly constrained by memory bandwidth and energy efficiency. Photonic computing offers a promising alternative due to its ultra-high bandwidth, massive parallelism, and low power dissipation. However, most existing photonic systems are limited to low-precision computation because of analog optical modulation constraints and noise accumulation, which restricts their applicability in precision-critical workloads. To address this limitation, we propose LightMat-HP, a hybrid photonic-electronic computing system that enables end-to-end acceleration of general matrix multiplication with configurable computational precision. LightMat-HP adopts block floating-point (BFP) arithmetic to reduce computational complexity while enabling flexible precision-performance tradeoffs. To overcome the precision limitations of photonic devices, we propose a slicing-based photonic multiplication scheme that exploits the high accuracy of low bit-width photonic multiplication in combination with digital accumulation to achieve high-precision mantissa multiplication. A tile-based matrix multiplication dataflow is further designed to support matrices of arbitrary sizes. We experimentally validate LightMat-HP on a photonic computing prototype and evaluate its performance through large-scale simulations. The results demonstrate that LightMat-HP outperforms FPGA, GPU, and a state-of-the-art photonic accelerator across throughput, latency, and energy efficiency, particularly for small- and medium-sized matrix multiplications, owing to its highly parallel photonic architecture, efficient data movement, and slice-based BFP arithmetic.

Problem

• matrix multiplication
• photonic computing
• computational precision
• analog optical modulation
• noise accumulation

Innovation

• photonic-electronic hybrid computing
• configurable precision
• block floating-point arithmetic
• slicing-based photonic multiplication
• matrix multiplication acceleration