基于光子混合专家网络的光子计算芯片、系统及方法
By using a photonic computing chip based on a photonic hybrid expert network and employing a passive optical diffraction structure for efficient computation, the computational bottleneck of traditional electronic computing platforms is solved, achieving low power consumption, high parallelism, and light-speed multi-task parallel processing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TSINGHUA UNIVERSITY
- Filing Date
- 2025-08-21
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional electronic computing platforms suffer from high power consumption, low efficiency, and architectural bottlenecks when faced with large-scale AI computing demands, making it difficult to meet the requirements of high parallelism and high throughput computing.
A photonic computing chip based on a photonic hybrid expert network is adopted, which utilizes a passive optical diffraction structure for efficient computing. Multi-task parallel processing is achieved through on-chip input waveguides, amplitude modulation interferometers, parallel photonic expert networks, and photodetectors, and data processing is combined with a back-end shared digital network.
It achieves low power consumption, high parallelism, and light-speed processing, significantly improving computing efficiency and capacity, and overcoming the energy efficiency and bandwidth limitations of traditional electronic computing.
Smart Images
Figure CN121189406B_ABST