一种近存计算芯片架构及方法
By utilizing arithmetic logic units within the near-memory computing group for data computation in the near-memory computing chip architecture, the memory wall problem is solved, chip computing power is improved, and bandwidth and power consumption are reduced.
CN122086839BActive Publication Date: 2026-07-17METAX INTEGRATED CIRCUITS (SHANGHAI) CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- METAX INTEGRATED CIRCUITS (SHANGHAI) CO LTD
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
The memory wall problem prevents further improvements in chip computing power, as memory bandwidth limits the match between processor and data transfer speed.
Method used
It adopts a near-memory computing chip architecture, including computing modules, multi-level caches and high-bandwidth memory. By implementing data computation in the near-memory computing group, it reduces cross-level data movement and uses the arithmetic logic units in the near-memory computing group for computation.
Benefits of technology
It alleviated the memory wall problem, improved chip computing power, and reduced memory bandwidth usage and power consumption.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122086839B_ABST
Abstract
本申请涉及芯片设计技术领域，特别是涉及一种近存计算芯片架构及方法。其包括：计算模块，所述计算模块包括若干计算核心，每个计算核心包括寄存器、算术逻辑单元和一级缓存；二级缓存，所述二级缓存与所述计算模块连接；高带宽内存，所述高带宽内存包括若干高带宽内存裸片，每个高带宽内存裸片与所述二级缓存连接；所述一级缓存、二级缓存和每个高带宽内存裸片均包括若干近存计算组，每个近存计算组包含算术逻辑单元、互联模块和若干存储体，每一近存计算组中的存储体均通过互联模块与该近存计算组中的算术逻辑单元相连。本发明能够缓解存储墙的问题，提升芯片算力。
Need to check novelty before this filing date? Find Prior Art