Presentation
3 October 2024 ChatGPT at the speed of light: Monolithic photonic-electronic linear-algebra accelerators for large language models
Tzu-Chien Hsueh, Yeshaiahu Fainman, Bill Lin
Author Affiliations +
Abstract
This paper proposed to adopt advanced monolithic silicon-photonics integrated-circuits manufacturing capabilities to achieve a system-on-chip photonic-electronic linear-algebra accelerator with the features of broadband incoherent photo-detections and high-dimensional operations of consecutive matrix-matrix multiplications for enabling substantial leaps in computation density and energy efficiency with practical considerations of power/area overhead due to photonic-electronic on-chip conversions, integrations and calibrations through holistic co-design approaches to support attention-head mechanism based deep-learning neural networks used in Large Language Models and other emergent applications.
Conference Presentation
© (2024) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Tzu-Chien Hsueh, Yeshaiahu Fainman, and Bill Lin "ChatGPT at the speed of light: Monolithic photonic-electronic linear-algebra accelerators for large language models", Proc. SPIE PC13113, Photonic Computing: From Materials and Devices to Systems and Applications, PC131130I (3 October 2024); https://doi.org/10.1117/12.3027309
Advertisement
Advertisement
KEYWORDS
Broadband telecommunications

Calibration

Energy conversion efficiency

Energy efficiency

Manufacturing

Neural networks

Silicon

Back to Top