Segmenting Large Bit Width Data for Limited-Width Processors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Processors with limited bit width struggle to efficiently process data of larger bit widths, leading to restricted computing capabilities and inefficiencies in operations such as neural network computations.
Innovation Solution
The approach involves representing large bit width data using smaller bit width components, allowing these components to perform operations and then combining the results to achieve the final outcome, thereby overcoming the limitations of the processor's bit width.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a processor with limited bit width is used to process large bit width data, then the processor can be used with lower computing costs and overheads, but the processing capability and computing efficiency are restricted
Solution Approach 1:
The patent divides large bit width data into multiple small bit width components. For example, a 32-bit data is segmented into multiple 8-bit components. This segmentation allows the limited-bit-width processor to process each component separately and combine the results, thereby maintaining compatibility with low-cost processors while achieving the capability to handle large bit width data through systematic division and recombination of data segments
2Ease of manufacture
If large bit width data is processed directly by a limited bit width processor, then computing costs are reduced, but the device complexity increases due to multiple processing steps
Solution Approach 1:
The patent segments large bit width data into multiple small bit width components that can be processed by the limited-bit-width processor. This segmentation approach maintains processor simplicity while enabling complex data handling through systematic division of the data processing task into manageable segments that can be handled by the existing processor architecture
Solution Approach 2:
The patent performs preliminary actions by pre-segmenting the large bit width data into smaller components before processing. This preliminary segmentation prepares the data in a format suitable for the limited-bit-width processor, reducing the complexity during the actual processing phase by having the data already organized into processable units
3Productivity
If small bit width components are used to represent large bit width data, then the processor can handle larger data with full computing power utilization, but the operation complexity increases due to multiple intermediate results
Solution Approach 1:
The patent segments large bit width data into multiple small bit width components, allowing the processor to utilize its full computing power by processing each component separately. The segmentation enables parallel or sequential processing of components, maximizing the processor's computational capabilities while handling data that exceeds its native bit width through systematic division and recombination
4Measurement precision
If multiple small bit width components are used to represent large bit width data, then computing precision is maintained, but the loss of time increases due to multiple processing steps
Solution Approach 1:
The patent segments large bit width data into multiple small bit width components, maintaining computing precision by preserving the integrity of each data segment. The segmentation approach ensures that each component is processed with the processor's full precision capabilities, and the final result is reconstructed by combining the processed components, thereby maintaining overall computational accuracy
5Ease of manufacture
If large bit width data is processed by a limited bit width processor, then hardware costs are reduced, but the manufacturing precision of the computing system decreases
Solution Approach 1:
The patent segments large bit width data into multiple small bit width components, allowing the limited-bit-width processor to handle each segment with its full manufacturing precision. This segmentation approach enables the use of lower-cost, limited-bit-width hardware while maintaining computing precision through systematic processing and recombination of data segments, effectively decoupling hardware cost from processing capability
Data Source
AI summary
The present disclosure discloses a computing apparatus, an integrated circuit chip, a board card, a device, and a method. The computing apparatus may be included in a combined processing apparatus. The combined processing apparatus may further include an interface apparatus 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 computing apparatus and other processing apparatus, respectively. The storage apparatus is used to store data of the computing apparatus and other processing apparatus. A solution of the present disclosure may use at least two pieces of small bit width data representing large bit width data to perform operation processing, so that processing capacity of a processor is not influenced by a bit width of the processor.