一种基于通用寄存器和共享内存的运算系统
By introducing a selector between shared memory and general-purpose registers and connecting it to the arithmetic unit, the problem that shared memory data cannot directly participate in operations is solved, resulting in more efficient operations and less instruction cache pressure, thus improving program performance.
CN121858164BActive 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
- 2026-03-17
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing technologies, data in shared memory cannot be directly used in computation, resulting in low computational efficiency, high instruction cache pressure, and wasted general-purpose register space.
Method used
By introducing a selector between shared memory and general-purpose registers and connecting it to the arithmetic unit, data in shared memory can be directly used in computation, simplifying the data path and reducing the number of instructions.
Benefits of technology
It improves computational efficiency, reduces instruction cache pressure, and saves general-purpose register space for other tasks, thereby enhancing program performance.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN121858164B_ABST
Abstract
本申请涉及集成电路设计技术领域，特别是涉及一种基于通用寄存器和共享内存的运算系统，该系统将共享内存和通用寄存器通过选择器与运算单元连接，使得共享内存中的数据可以直接参与运算单元的运算，简化了共享内存中的数据参与运算单元运算的通路，而且可以减少指令数量，减小指令缓存压力，提高运算效率，另外，所节省的通用寄存器的空间可以应用于其他任务，潜在提高了程序性能。
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Visual processing device based on multi-layer parallel processing
CN103020890A
Processor cluster structure based on shared register file and global synchronization module
CN103984672A
Method and system for achieving TensorCore convolution calculation in TVM
CN111949932A