Supercharge Your Innovation With Domain-Expert AI Agents!

A large-scale chip-on-chip interconnection method and routing algorithm for realizing interconnection structure

A chip-on-chip, large-scale technology, applied in digital transmission systems, electrical components, transmission systems, etc., can solve problems such as inability to adapt to localized computing of nodes and localized data storage, and achieve the effect of speeding up communication efficiency

Active Publication Date: 2017-11-17
THE PLA INFORMATION ENG UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] A cross-domain direct connection method for large-scale on-chip chip interconnection. In the design of chip-on-chip network structure with network interconnection structure, all nodes in the current on-chip network structure are peer-to-peer connected together, which cannot adapt to node localization calculation and data localization. To solve the problem of actual demand for storage, take the following steps:

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 large-scale chip-on-chip interconnection method and routing algorithm for realizing interconnection structure
  • A large-scale chip-on-chip interconnection method and routing algorithm for realizing interconnection structure
  • A large-scale chip-on-chip interconnection method and routing algorithm for realizing interconnection structure

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0021] Example 1: see figure 2 , The cross-domain direct connection large-scale chip-on-chip interconnection method of the present invention, in the chip-on-chip network structure design with a network interconnection structure, is aimed at the peer-to-peer connection of nodes in the current on-chip network structure, which cannot adapt to node localized calculations and data The actual demand for localized storage,

[0022] 1) At the peer node location, instantiate a local subnet and add the local computing node (IP) connected to the local subnet as a local computing domain;

[0023] 2) At the same time, between adjacent local computing domains, they are directly connected through the routing nodes of the local subnet to build the main network of the on-chip network structure to speed up the communication efficiency between different local computing domains.

Embodiment 2

[0024] Example 2: see figure 1 , figure 2 Compared with the embodiment, the cross-domain direct connection large-scale chip-on-chip interconnection method of this embodiment furthermore includes a local computing domain resource (P) and adjacent routers for each local subnet at the peer node location (R), each local computing domain resource is connected to a backbone router through a network interface, where the local computing domain resources are processor cores, memory or user-defined hardware modules, and the backbone router is adjacent to the main network The router (R) is directly connected and connected to the four subnet routers of the local subnet at the same time.

Embodiment 3

[0025] Embodiment 3: This embodiment takes the simplest network topology 4*4 mesh network as an example to illustrate the design idea of ​​the cross-domain direct connection large-scale chip-on-chip interconnect structure of the present invention. Each node in the network is connected to a resource and its neighboring routers. Here is different from other on-chip network structures in that each resource is not connected to the corresponding router through a network interface (NI). This router is directly embedded in the resource and connected to the internal router in the resource. Here we are for convenience Observe and draw it outside. The so-called resources here are chips with a network interconnection structure. The router and the router are connected by a pair of input and output channels. The channel is composed of two unidirectional point-to-point buses.

[0026] For the internal structure of resources in the main network, we also use the simplest network topology in t...

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 method for realizing a chip interconnection structure. A cross-domain direct connection method for large-scale on-chip chip interconnection. For the current on-chip network structure, 1) instantiate a local subnet at the peer node position, and add the local computing nodes connected to the local subnet as a local computing domain ; 2) Between the local computing domains in adjacent locations, the routing nodes of the local subnet are directly connected to construct the main network of the network-on-chip structure. The routing algorithm of the cross-domain direct connection large-scale chip-on-chip interconnection structure, for each local computing domain resource in the chip-on-chip interconnection structure, when the data packet is sent from the source address, it is first judged whether its destination address is in the subnet , if so, use the routing algorithm suitable for the subnet to deliver the data packet; if it is not in the subnet, first send the data packet to the router connected to the external network, and then read the main network address in the destination address in the router, According to the routing algorithm of the main network, the data packet is sent to the subnet corresponding to the destination address; then according to the destination address of the subnet in the data packet, the data packet is sent to the destination address through the routing algorithm of the subnet.

Description

Technical field [0001] The present invention relates to a chip interconnection structure, in particular to a large-scale chip on-chip interconnection structure based on the idea of ​​cross-domain direct connection and a routing algorithm for realizing the structure. Background technique [0002] A typical system-on-chip design often uses two communication strategies, on-chip bus and on-chip network. The main advantages of the on-chip bus are high flexibility, scalability, and low design cost. Generally, the time delay is shorter when the bandwidth requirement is lower; the disadvantage is that the long communication line brings certain energy consumption and limits the system clock rate. When there are more than two modules in the communication structure, scalability is reduced. [0003] The main advantage of the network-on-chip technology is that it can support concurrent communication between hardware modules, is more scalable, and can be used to support greater bandwidth. The ...

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 Patents(China)
IPC IPC(8): H04L12/721H04L12/931
Inventor 韩国栋张兴明张效军刘勤让张帆贺涛沈剑良曾威朱珂
Owner THE PLA INFORMATION ENG UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More