A vectorized quick sorting calculation method
Through the vectorized quick sorting calculation method, the data segmentation and parallel processing is used to process data segmentation and parallel, which solves the problems of high time complexity and low vectorization of the existing sorting algorithm, and realizes efficient sorting calculation.
CN119739425APending Publication Date: 2025-04-01JIANGSU HUACHUANG MICROSYSTEM CO LTD
0 Cites 0 Cited by
Patent Information
- Application Number
- CN202411852974.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-16
- Publication Date
- 2025-04-01
AI Technical Summary
Technical Problem
When existing sorting algorithms process large amounts of data, the time complexity is high and difficult to implement vectorization processing, resulting in low computational efficiency and poor performance.
Method used
A vectorized quick sorting calculation method is proposed, which can process data in segments and use multiple vector registers to perform parallel calculations, optimize the traversal process, reduce repeated data transfer, and realize parallel calculation of single instruction and multiple data streams.
Benefits of technology
It effectively improves the performance of the sorting algorithm, reduces time complexity, enhances computing efficiency, and realizes the universality and portability of cross-platform chip algorithms.
✦ Generated by Eureka AI based on patent content.
Abstract
The present invention discloses a vectorized quick sorting calculation method, comprising the steps of: S1, initial setting and segmentation; S2, judging whether the loop variable is less than N, if not, go to S5; if so, calculate the current segment address; S3, basic sorting; S4, updating the loop variable, go to S2; S5, judging whether there is tail data, if not, go to S6; if so, perform basic sorting of tail data; S6, judging whether the segment length sn is less than the data length n, if not, go to S9; if so, enter S7; S7, perform s-way merge sorting; S8, set the segment length sn*=s, go to S6; S9, judging whether the current destination address is the output vector address, if not, copy the result to the output vector; if so, sorting ends. The present invention solves the problems of repeated data movement, high time complexity and low vectorization degree in sorting calculation, improves performance and is applicable to a wider range of platforms and chips.
Need to check novelty before this filing date? Find Prior Art