Request processing method, apparatus, device and medium

Abstract

The present application discloses a request processing method, device, device and medium, and relates to the field of management software, including: obtaining an address of a current pending request from a local first-in-first-out queue; Obtain the current pending request from the buffer; parse the current pending request, and obtain the corresponding current pending source data and the serial address item used to record the address of the next pending request based on the parsing result; Process the data to obtain the currently processed target data, write the current processed target data into the preset data buffer, and determine whether the serial address item is empty; if the serial address item is empty, notify the host to read from the preset data buffer Get all the processed target data, if the serial address item is not empty, send the address recorded in the serial address item to the FIFO queue of the next accelerator card for storage. To achieve the purpose of reducing the CPU burden and improving the efficiency of request processing.

Description

technical field [0001] The present invention relates to the field of management software, in particular to a request processing method, apparatus, device and medium. Background technique [0002] In recent years, with the development of emerging technologies such as mobile Internet and artificial intelligence, the scale of big data and the scale of Internet applications are growing exponentially. The business that data centers need to handle has also grown, and the computing power of general-purpose processors used in data centers has reached a bottleneck. In order to solve this problem, various PCIe (peripheral componentinterconnect express, high-speed serial computers) based on FPGA (FieldProgrammable Gate Array, Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit, special integrated circuit) dedicated computing accelerated PCIe (peripheral componentinterconnect express) Expansion bus standard) boards came into being, such as compression and deco...

AI Technical Summary

Problems solved by technology

However, since different accelerator cards use different driver interfaces, in the prior art, when multiple accelerated calculations are required, each step of calculation processing needs to be scheduled by the host, which undoubtedly increases the CPU (central processing unit, that is, the central processing unit). ) burden and reduce processing efficiency

Images