Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Multi-GPU (graphic processing unit) interconnection system structure in heterogeneous system

An architecture and heterogeneous system technology, applied in the field of computer communication, to achieve the effect of improving system performance and supporting scalability

Active Publication Date: 2012-07-04
THE PLA INFORMATION ENG UNIV +1
View PDF4 Cites 48 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The technical problem to be solved by the present invention is: to overcome the defects of the existing technology, in order to solve the problem of high-speed interconnection of multiple GPUs in the existing heterogeneous system, and to support the requirements of scalability and interface flexibility, the present invention provides a multi-level PCI -E exchange structure, and realize transparent transmission, memory sharing and other functions

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
  • Multi-GPU (graphic processing unit) interconnection system structure in heterogeneous system
  • Multi-GPU (graphic processing unit) interconnection system structure in heterogeneous system
  • Multi-GPU (graphic processing unit) interconnection system structure in heterogeneous system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] Example 1: see figure 1 , figure 2 , The multi-GPU interconnection architecture in the heterogeneous system of the present invention

[0035] In a hybrid and heterogeneous high-performance computer system that combines multi-core accelerators (GPU) and multi-core general-purpose processors (CPU), multiple GPUs and multiple GPUs and CPU processors are used based on the PCI-E bus. The port switching chip performs multi-level interconnection to form a multi-level switching structure with reconfigurable external interfaces. The multi-level switching structure adopts a four-port switching chip. Every two GPUs are interconnected through a PCI-E interface of the four-port switching chip to establish an internal channel between dual GPU devices. At the same time, the upper switching chip provides a downstream port and the next Four-port switching chip communication connection forms a hierarchical switching structure.

[0036] The characteristics of GPU devices determine that the G...

Embodiment 2

[0037] Example 2: see figure 1 , figure 2 , image 3 . Tesla C1060 only provides PCI-E interface, because

[0038] This application PCI-E switching chip solves the problem of multi-GPU interconnection. However, the PCI-E switching chip only provides one uplink channel. In the heterogeneous system of the present invention, the multi-GPU interconnection architecture uses a four-port switching chip innovatively, and one of the downlink ports is used to communicate with other components. Inside the system, every two GPUs are connected to the PCI-E switch chip through the PCI-E interface. The upstream of the switch chip is used as the downstream input of another switch chip, and one of the downstream ports of the uppermost switch chip is used to communicate with other The components communicate. The system provides custom or IB interfaces for data exchange and interconnection with other external computing components. The Ethernet port is used for system management and maintenance. ...

Embodiment 3

[0044] Example three: see Figure 4 . The multi-GPU interconnection system structure in the heterogeneous system of the present invention is

[0045] Nvidia's TurboCache (TC) technology is also used in the heterogeneous system to build a software support environment for collaborative computing in heterogeneous systems. In view of the particularity and complexity brought by the hybrid system architecture, it fully taps general-purpose processors and special-purpose processors. The high-performance advantages of accelerators in dealing with specific problems, the construction of a heterogeneous processor collaborative computing environment based on the CPU multi-core architecture and the GPU many-core architecture, and the realization of a GPU-CPU high-performance cluster management system for the many-core architecture, Programming model and related software development environment. It can support the GPU to directly access the system memory without using the local memory on the...

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 relates to a GPU (graphic processing unit) hardware configuration management problem in a heterogeneous system in the technical field of computer communication, and in particular relates to a multi-GPU interconnection system structure in the heterogeneous system. In the multi-GPU interconnection system structure in the heterogeneous system, based on a mixed, heterogeneous and high-performance computer system which is combined by multi-core accelerators and a multi-core universal processor, multiple multi-core accelerators, as well as between multiple multi-core accelerators and the multi-core universal processor are subjected to multistage interconnection by adopting multi-port exchange chips based on PCI-E (peripheral component interconnect-express) buses, and a multistage exchange structure which is reconfigurable to an external interface is formed. The multi-GPU interconnection system structure in the heterogeneous system provided by the invention solves the high-speed interconnection problem between the multiple GPUs and between the multiple GPUs and the CPU in the traditional heterogeneous system, is a hardware system structure with flexibility and expandability, supports the expandability and interface flexibility and realizes transparent transmission and memory sharing.

Description

Technical field [0001] The present invention relates to the technical field of computer communication, in particular to the problem of GPU hardware configuration management in a heterogeneous system in the computer communication field, and in particular to a multi-GPU interconnection architecture in a heterogeneous system. Background technique [0002] GPU (Graphics Processing Unit), or "graphics processing unit". Early GPUs were only responsible for graphics rendering, and most of the processing was handed over to the CPU. [0003] In recent years, with the development of computer hardware technology, the improvement of processor and network performance, and the decline in prices, parallel computing has begun to shift from the traditional high-performance computer system to a scale composed of a group of high-performance nodes or workstations / PCs. It is the computing platform of the cluster. At the same time, the development speed of GPU in the field of graphics processing has f...

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): G06F15/163G06T1/00
Inventor 王磊周蓓李柏楠张凤娟郭绍忠
Owner THE PLA INFORMATION ENG UNIV
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
Eureka Blog
Learn More
PatSnap group products