A GPU card cluster configuration control system and method

A GPU card and configuration control technology, applied in the field of GPU card cluster configuration control system, can solve problems such as unfavorable product cost optimization, resource waste, resource redundancy, etc., and achieve the effects of convenient maintenance, saving development costs, and flexible application

Inactive Publication Date: 2019-05-10
ZHENGZHOU YUNHAI INFORMATION TECH CO LTD
View PDF5 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The design of the management module with dual control chips has resource redundancy, which w

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 GPU card cluster configuration control system and method
  • A GPU card cluster configuration control system and method
  • A GPU card cluster configuration control system and method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0028] Such as figure 2 As shown, the GPU node includes a BMC, a management network interface RJ45 respectively connected to the BMC, and at least one GPU extension chip. The I2C interface of the extension chip is connected to the I2C interface of the BMC; the extension chip is connected to two GPU cards respectively. Use the I2C bus to control the configuration and replace the original PCIE link, thereby releasing CPU resources from GPU nodes, saving resources and reducing costs.

Embodiment 2

[0030] Such as image 3 As shown, this embodiment is a GPU cluster construction situation including multiple GPU nodes. The network interface RJ45 of the GPU node is connected to the network interface of the switch module, and the network interface of the switch module controlling the switchboard is connected to the network interface of the switch module. When the GPU cluster is built, each GPU node is connected to the switch through the management network interface RJ45 of the BMC, so that the management configuration of all GPU nodes is connected to the same network, and the control switchboard is connected to the switch to control the configuration of each GPU node. The BMC of each GPU node is connected through a switch to manage the allocation of each GPU card in the entire GPU cluster.

[0031] Such as Figure 4 As shown, a GPU card cluster configuration control method includes the following steps:

[0032] S1. Log in to the webpage management interface of the BMC contr...

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

The invention discloses a GPU card cluster configuration control system. The system comprises at least one GPU node and a network expansion unit connected with a network communication end of the GPU node. The GPU node comprises a BMC control module, a management network interface module and at least one GPU card expansion module, wherein the management network interface module and the at least oneGPU card expansion module are respectively connected with the BMC control module, and the GPU card expansion module is connected with a GPU card. The invention further provides a GPU card cluster configuration control method. By integrating the management resources, the redundant resources in the GPU nodes are integrated, the system resources are reasonably utilized, and the development cost is saved.

Description

technical field [0001] The invention relates to the technical field of fusion architecture, in particular to a GPU card cluster configuration control system and method. Background technique [0002] At present, AI technology is developing rapidly, and the computer architecture with high computing performance has also experienced an unprecedented surge in research and development. At present, the GPU cards with high computing performance released by NVIDIA occupy a leading position in computing performance. Parallel design of multiple GPU cards to form a GPU card computing cluster, combined with computing servers, has become a computing system that continuously improves computing performance in the industry. [0003] The GPU card takes the GPU node as the deployment unit, and integrates a large-scale GPU cluster by integrating multiple GPU nodes. In order to realize the resource pooling of GPU cards, GPU cards can be allocated to any upstream host. Each GPU card node is equ...

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
IPC IPC(8): G06F9/4401G06F9/50G06F13/42
Inventor 王玲燕
Owner ZHENGZHOU YUNHAI INFORMATION TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap