Full-link resource scheduling method and device for PS-PL-GPU three-level heterogeneous system

CN121144033APending Publication Date: 2025-12-16CHENGDU NOSTIAN TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511296419.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-09-11
Publication Date
2025-12-16

Smart Images

  • Figure CN121144033A_ABST
    Figure CN121144033A_ABST
Patent Text Reader

Abstract

The invention discloses a full-link resource scheduling method and device oriented to a PS-PL-GPU three-level heterogeneous system, and the method comprises the steps: constructing a three-level heterogeneous basic operation model comprising a PS end, a PL end and a GPU end, carrying out the pre-distribution of processing tasks for the PS end, the PL end and the GPU end, and building a hardware buffering module for data buffering at the PL end; and monitoring operation state indexes of the PS end, the PL end and the GPU end in real time, and when the operation state index of any end reaches a preset standard, triggering to execute a dynamic optimization strategy corresponding to the preset standard, and adjusting a data processing or transmission process. According to the invention, the overall efficiency of data processing is improved.
Need to check novelty before this filing date? Find Prior Art