Unlock instant, AI-driven research and patent intelligence for your innovation.

A method and a device for testing a GPU-BOX system

A technology for system testing and testing indicators, applied in faulty hardware testing methods, detection of faulty computer hardware, error detection/correction, etc., can solve problems such as comprehensive and effective evaluation, inability to achieve GPU system performance, and single testing

Active Publication Date: 2019-04-26
ZHENGZHOU YUNHAI INFORMATION TECH CO LTD
View PDF11 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, for the current deep learning field, the evaluation test of the GPU system lacks some effective methods and means. Through a single test of the server GPU-BOX system, it is impossible to achieve a comprehensive and effective evaluation of the performance of the GPU system.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method and a device for testing a GPU-BOX system
  • A method and a device for testing a GPU-BOX system
  • A method and a device for testing a GPU-BOX system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] Such as figure 1 As shown, the present invention provides a kind of method for GPU-BOX system test, comprising:

[0044] S1, configure the GPU-BOX system, check the GPU bandwidth, and obtain bandwidth information;

[0045] S2, check the number of GPUs and the version of Vbios, and obtain the test index parameters in the GPU-BOX system performance test process;

[0046] S3, connect the BOX to the server, pressurize the system on the server side, and test the overall power consumption of the GPU-BOX system.

Embodiment 2

[0048] Such as figure 2 As shown, in step S1, the GPU-BOX system is configured, the GPU bandwidth check is performed, and the bandwidth information obtained specifically includes:

[0049] S11, modify the / boot / grub.conf configuration file under the Red Hat system, and disable the graphical interface;

[0050] S12, use the loop lspci command to capture all the GPUs in the BOX and obtain bandwidth information;

[0051] S13, check whether the bandwidth information is x16, if the judgment result is yes, then execute step S14, if the judgment result is no, then execute step S15;

[0052] S14, proceed to the next step of performance testing;

[0053] S15, check the connection between the PICE slot and the GPU, and re-execute step S12 after confirming that its bandwidth is x16.

[0054] Wherein, in step S11, disabling the graphical interface is specifically modifying the kernel parameter "intel_iommu" from on to off, and the specific command is: "intel_iommu=on amd_iommu=on" is ch...

Embodiment 3

[0056] Such as image 3 As shown, in step S2, check the number of GPUs and the version of Vbios, and obtain the test index parameters in the GPU-BOX system performance test process specifically include:

[0057] S21, check the number of GPUs and the version of Vbios by looping the lspci command;

[0058] S22, obtain the test index parameters in the end-to-end test process in the GPU-BOX system performance test, and check whether the test index parameters meet the stress test standard in the performance test, if the judgment result is yes, then perform step S23; if the judgment result If no, execute step S24;

[0059] S23, continue to perform the follow-up stress test;

[0060] S24, indicating that there is a problem in the hardware system itself, and the test is ended.

[0061] Before step S21 is carried out, the test tool NVQual also needs to be installed, and the GPU-BOX system is tested end-to-end using the CUDA test tool in the test tool NVQual (referring to the GPUs in...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

In order to solve the problems in the prior art, the present invention proposes a method for testing a GPU-BOX system, including: configuring a GPU-BOX system to perform GPU bandwidth check; performing performance test on a GPU-BOX system; carrying out the power consumption test of the whole machine on a GPU-BOX system. a device for testing the GPU-BOX system is also proposed. The technical solution of the present invention compensates and perfects the performance test of the server GPU-BOX system, and greatly improves the test efficiency and stability of the GPU-BOX system. an efficient testverification and evaluation system is provided for the GPU-BOX system, which brings a new solution to the GPU-BOX system test in the artificial intelligence field, which can comprehensively, systematically and effectively evaluate the GPU system.

Description

technical field [0001] The invention relates to the field of system testing, in particular to a GPU-BOX system testing method and device. Background technique [0002] Artificial intelligence is playing an increasingly important role in the field of online services in the current world. Within companies such as Google, Facebook, Microsoft, and Baidu, GPUs play a huge role in the field of "deep learning" because GPUs can process a large number of applications in parallel. Trivia. Deep learning relies on a neural network—a network highly similar to the nerves of the human brain—and the purpose of this network is to analyze massive amounts of data at high speed. And compared to the CPU, another advantage of the GPU is that its energy demand is much lower than that of the CPU. However, so far, when these companies launch deep learning services, it is still the CPU in the data center system that drives the application. Current deep learning applications still rely on CPUs beca...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F11/22
CPCG06F11/2236G06F11/2273
Inventor 白云峰
Owner ZHENGZHOU YUNHAI INFORMATION TECH CO LTD