Processing Circuit Array for Flexible AI Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing instruction sets in computing systems, particularly in the AI field, face limitations in flexibility, leading to poor performance in terms of throughput, execution speed, execution efficiency, and power consumption due to hardware architecture constraints.
Innovation Solution
A hardware architecture featuring a processing circuit array formed by connecting multiple processing circuits in one-dimensional or multi-dimensional arrays, allowing for multi-thread operations and flexible configuration to perform multiple operations with reduced computing overheads and increased efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing instruction sets are used in hardware architecture, then computing operations can be performed, but flexibility is limited and processing performance is poor
Solution Approach 1:
The patent divides the processing circuit into multiple processing circuit sub-arrays that can be independently configured and controlled. Each sub-array can be dynamically assigned to different operations based on the computing instruction, enabling flexible adaptation to various computational tasks while maintaining high processing performance through parallel execution.
Solution Approach 2:
The patent implements dynamic configuration of the processing circuit array where sub-arrays can be reconfigured based on operational requirements. The system can dynamically adjust the number of active sub-arrays, their arrangement, and their assigned operations to optimize both flexibility and processing performance for different computational workloads.
2Productivity
If multiple operations require multiple instructions, then computing tasks can be completed, but throughput of on-chip I/O data increases
Solution Approach 1:
The patent merges multiple operations into a single computing instruction by parsing the instruction into multiple operation instructions that are executed in parallel across different processing circuit sub-arrays. This consolidation reduces the number of separate instructions needed, decreases on-chip I/O data throughput requirements, and improves execution efficiency through simultaneous operation execution.
Solution Approach 2:
The patent enables continuous execution of multiple operations through parallel processing circuits that operate simultaneously. By distributing operations across multiple sub-arrays that work in parallel, the system maintains continuous productive action without idle waiting periods, thereby improving execution efficiency while reducing the cumulative data throughput burden compared to sequential instruction execution.
3Speed
If existing instructions are used, then computing operations can be performed, but execution speed and power consumption have room for improvement
Solution Approach 1:
The patent segments the computing task across multiple processing circuit sub-arrays that can execute operations in parallel. This segmentation enables faster execution speed by performing multiple operations simultaneously rather than sequentially, while power consumption is optimized by activating only the necessary number of sub-arrays for each specific computing task, avoiding the energy waste of keeping all circuits active.
Solution Approach 2:
The patent applies partial action by activating only the required number of processing circuit sub-arrays based on the complexity and requirements of the computing instruction. Instead of running all possible operations at full capacity, the system dynamically adjusts the level of computational action to match the actual task requirements, thereby achieving fast execution speeds while optimizing power consumption to avoid excessive energy usage.
Data Source
AI summary
A computing apparatus may be included in a combined processing apparatus. The combined processing apparatus may further include a general interconnection interface and other processing apparatus. The computing apparatus interacts with other processing apparatus to jointly complete a computing operation specified by a user. The combined processing apparatus may further include a storage apparatus. The storage apparatus is connected to the apparatus and other processing apparatus respectively. The storage apparatus is used to store data of the apparatus and other processing apparatus. Efficiency of various operations in data processing fields including, for example, an artificial intelligence field may be improved so that overall overheads and costs of the operations can be reduced.


