Optical in-memory computing systems

Optical in-memory computing systems, employing spatial light modulators and hyperspectral operations, address the inefficiencies of traditional computing architectures in matrix-vector multiplication, achieving high computational throughput and energy efficiency.

WO2025080846A1 Publication Date: 2025-04-17NTT RESEARCH INC
7 Cites 0 Cited by

Patent Information

Application Number
PCT/US2024/050792
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-10-14
Filing Date
2024-10-10
Publication Date
2025-04-17

AI Technical Summary

Technical Problem

Traditional von Neumann computing architectures face challenges in performing extensive matrix-vector multiplication operations efficiently and energy-effectively, particularly in deep learning and optimization tasks.

Method used

The development of optical in-memory computing systems that utilize spatial light modulators as programmable optical memory, integrating hyperspectral operations with space and frequency multiplexing to enhance computational throughput and energy efficiency.

Benefits of technology

These systems demonstrate the ability to perform multiply-accumulate operations with higher than 4-bit precision in matrix-vector and matrix-matrix multiplications, showcasing potential for deep learning and optimization tasks while offering extraordinary modularity, scalability, and programmability.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

According to some embodiments of the present disclosure, optical computing systems are disclosed that employ optical sources, such as optical frequency combs, and optical memories to perform computations such as Matrix-Vector Multiplication (MVM) operations or Matrix-Matrix Multiplication (MMM) operations. According to some embodiments of the present disclosure, optical computing systems are disclosed that employ optical sources, such as optical frequency combs, optical memories, and components configured to perform hyperspectral operations to increase throughput.
Need to check novelty before this filing date? Find Prior Art