GPU mode switching stability test method and system, terminal and storage medium

Abstract

The invention provides a GPU mode switching stability test method and system, a terminal and a storage medium. The method comprises the steps of setting a bandwidth reference of a GPU at a speed reduction rate and a bandwidth reference of the GPU at a maximum rate; reducing the sensitivity of the CPU end to PCIE error reporting by setting the PCIE error reporting register to be zero; limiting theGPU-PCIE link rate to be the speed reduction rate by setting the limiting rate of a CPU register; pressurizing the GPU, comparing the pressurized bandwidth of the GPU with the bandwidth reference at the corresponding rate, and if the pressurized bandwidth of the GPU is inconsistent with the bandwidth reference at the corresponding rate, outputting that the test fails; and cancelling the rate setting of the CPU register, periodically pressurizing the GPU, comparing the bandwidth of the GPU in the periodic pressurizing process with the corresponding reference bandwidth, and if the bandwidths areinconsistent, outputting that the test fails. According to the method, the PCIE rate can be switched in an active mode and a passive mode to carry out GPU compatibility stability testing. And the test coverage of GPU and machine compatibility is perfected. A verification mode is provided for small probability events, and the product quality is guaranteed.

Description

technical field [0001] The invention relates to the technical field of server testing, in particular to a GPU mode switching stability testing method, system, terminal and storage medium. Background technique [0002] GPU (Graphics Processing Unit) - Graphics processor, also known as visual processor, is a microprocessor that specializes in image and graphics-related operations. GPU reduces the dependence of graphics card on CPU during 3D graphics processing, and shares the Part of the original CPU work. With the advent of Internet 2.0, each user of various video graphics dissemination APPs (Weibo, WeChat, Douyin, Volcano Video, etc.) is an author, and various graphic video DIY needs a lot of image processing. Therefore, in recent years, the utilization and development of GPU has become more extensive. The stable operation of the GPUs of various APP manufacturers is related to the customer satisfaction of APP users. Stability testing for GPUs becomes extremely important. ...

AI Technical Summary

Problems solved by technology

[0004] However, in the development process of the commonly used GPU BOX, due to the matching retimer card (its function is to correct the signal during long-distance transmission, etc.), the VDD18 power supply has a voltage jump of 20-30mV when the rate is switched, and the Speed ​​is performed on the corresponding PCIe link. During the change process, a PCIE error will be reported probabilistically, causing the system to crash and restart

Therefore, it is very difficult to realize the stability test under the scenario of frequent rate switching

Images