A GPU server request processing method, device, equipment and storage medium

Abstract

This application discloses a GPU server request processing method, apparatus, device, and storage medium, relating to the field of GPU server resource scheduling technology. The method includes: receiving a request to be processed; acquiring a first feature dataset of the request to be processed and a second feature dataset of the GPU server; determining a type matching degree based on first type data in the first feature dataset and second type data in the second feature dataset, determining a resource matching degree based on first resource data and second resource data, and determining a constraint matching degree based on first constraint data and second constraint data; determining a comprehensive matching degree based on the three matching degrees; if there exists a first set of GPU processors with a comprehensive matching degree greater than or equal to a matching degree threshold, calculating the reliability of each GPU processing the request, and determining the target GPU processor with the highest reliability; and allocating the request to be processed to the target GPU processor for processing. This method can improve processing efficiency.

Images