Data-Level Pipeline for Multi-Storage Parallelism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pipeline technologies in vector computing primarily focus on pipelining between computing and caching units, failing to fully optimize the efficiency and performance across all levels of storage and computing units.
Innovation Solution
The implementation of a data-level pipeline that extends pipelining across different levels of storage units, including first- and second-level storage units and processing units, through reasonable space division, enabling multi-level pipelining and parallel processing across all units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If pipeline technology is implemented between computing and caching units, then parallel processing is achieved and acceleration ratio reaches up to 50%, but the efficiency improvement and optimization of pipeline technology cannot be fully tapped
Solution Approach 1:
The second-level storage unit is divided into a plurality of storage areas to enable independent parallel operations. This segmentation allows multiple data blocks to be processed simultaneously at different pipeline stages, expanding the coverage of pipeline optimization beyond the traditional computing-caching interface to include multi-level storage parallelism.
Solution Approach 2:
The patent extends pipeline technology from a single dimension (computing-caching parallelism) to multiple dimensions by incorporating multi-level storage units (first-level and second-level) with their respective storage areas. This dimensional expansion enables parallelism across storage hierarchy levels while maintaining computing-caching parallelism, fully tapping pipeline optimization potential.
2Productivity
If the second-level storage unit is divided into multiple storage areas for multi-level pipelining, then parallel processing across all units is enabled, but device complexity increases
Solution Approach 1:
The second-level storage unit is divided into a plurality of storage areas, each capable of independent operation. This segmentation enables multiple data blocks to be processed in parallel at different pipeline stages, achieving multi-level parallelism while maintaining manageable complexity through modular organization of storage areas.
Solution Approach 2:
The storage areas in the second-level storage unit are designed to serve multiple functions: they can store data blocks for processing, hold intermediate results, and facilitate data transfer between different pipeline stages. This multi-functionality reduces the need for dedicated structures for each purpose, thereby controlling device complexity while enabling comprehensive parallel processing.
Data Source
AI summary
The present disclosure discloses a data processing apparatus, a data processing method, and related products. The data processing apparatus is used as a computing apparatus and is included in a combined processing apparatus. The combined processing apparatus further includes 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 further includes a storage apparatus. The storage apparatus is respectively connected to the computing apparatus and other processing apparatus and is used to store data of the computing apparatus and other processing apparatus. The solution of the present disclosure takes full advantage of parallelism among different storage units to improve utilization of each functional component.