Data processing methods, devices, chips, equipment and media

By breaking down the instructions to be processed into micro-instructions and executing them in cache units, the problem of efficiency degradation of vector processors in processing large amounts of data is solved, and more efficient data processing is achieved.

CN115033286BActive Publication Date: 2026-06-30KUNLUNXIN TECHNOLOGY (BEIJING) CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
KUNLUNXIN TECHNOLOGY (BEIJING) CO LTD
Filing Date
2022-06-13
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

When processing large amounts of data, existing vector processors require increased computational power from their execution units and longer processing times, leading to decreased processing efficiency.

Method used

The instruction to be processed is split into multiple micro-instructions and written into a preset cache unit. For any micro-instruction in the cache unit, the micro-instruction is processed based on the fact that the data in its corresponding source storage unit has been updated to the operand.

Benefits of technology

By splitting microinstructions, the amount of data processing per instruction is reduced, the computational resource requirements of the execution unit are decreased, execution order dependency is avoided, and overall processing efficiency is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115033286B_ABST
    Figure CN115033286B_ABST
Patent Text Reader

Abstract

This disclosure provides a data processing method, apparatus, chip, device, and medium, relating to the field of computer technology, and particularly to the field of chips. The implementation scheme includes: splitting a received instruction to be processed into multiple microinstructions, wherein each microinstruction is used to process data in a different source storage unit; writing the multiple microinstructions into a preset cache unit; and for any microinstruction in the cache unit, at least based on the fact that the data in the source storage unit corresponding to the microinstruction has been updated to the operand to be processed by the microinstruction, executing the microinstruction to process the operand.
Need to check novelty before this filing date? Find Prior Art