Server-based AI computing power intelligent allocation and management method and system

By extracting multi-dimensional allocation features from the AI ​​server and using a cross-fusion mechanism and digital AI model for simulation, the computing power parameters are dynamically adjusted, solving the problem of inaccurate computing power resource allocation in existing technologies and achieving efficient and reliable resource allocation.

CN122431882APending Publication Date: 2026-07-21NANJING TONGLIYU TECHNOLOGY CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANJING TONGLIYU TECHNOLOGY CO LTD
Filing Date
2026-04-28
Publication Date
2026-07-21

Smart Images

  • Figure CN122431882A_ABST
    Figure CN122431882A_ABST
Patent Text Reader

Abstract

The application relates to the technical field of computing resource allocation management, and discloses an AI computing power intelligent allocation and management method and system based on a server; the method comprises the following steps: S1, extracting multi-dimensional allocation features; S2, analyzing a target vector into computing power parameters; S3, compensating and optimizing the computing power parameters; and S4, allocating computing power resources of an AI server to a to-be-processed task; the application can pre-simulate the to-be-processed task and the computing power parameters on a virtual level through a digital AI model, and can dynamically adjust the computing power parameters of the to-be-processed task in a step-by-step compensation mode according to the simulation result, so that the initial allocation result of the computing power resources can be pre-performed, potential defects and deficiencies in the computing power allocation can be found, and reliable reference bases for subsequent compensation optimization and real allocation of the computing power resources are provided, the phenomenon of task execution failure caused by insufficient and biased computing power resource estimation is effectively avoided, and the reliability of the computing power resource allocation result is improved.
Need to check novelty before this filing date? Find Prior Art