Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Routing mechanism for heterogeneous many-core task scheduling based on artificial potential field

An artificial potential field method and task scheduling technology, applied in digital transmission systems, electrical components, transmission systems, etc., can solve the problem of not fully considering the task processing capacity of shared nodes, network congestion, and task scheduling algorithms that cannot achieve balanced use of shared resources and other issues to achieve the effect of preventing jitter, reducing congestion, and improving performance

Active Publication Date: 2015-08-05
SUZHOU POWERCORE TECH CO LTD
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This task scheduling method does not fully consider the task processing capabilities of the shared nodes. Once the source node visits a busy shared node, it will cause network congestion.
It can be seen that in the heterogeneous many-core structure, the task scheduling algorithm cannot achieve the balanced use of shared resources.

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
  • Routing mechanism for heterogeneous many-core task scheduling based on artificial potential field
  • Routing mechanism for heterogeneous many-core task scheduling based on artificial potential field
  • Routing mechanism for heterogeneous many-core task scheduling based on artificial potential field

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0036] A routing mechanism for heterogeneous many-core task scheduling based on the artificial potential field method described in the present invention, this mechanism is mainly used in heterogeneous many-core structures, NOC (Network-on-Chip, network on chip) Internet dynamic allocation common How to select shared processing cores (heterocores) for processing cores (common cores), such as image 3 As shown, it includes the following steps:

[0037] S1: The present invention introduces the artificial potential field method, and uses each shared processing core as a gravitational source to establish a gravitational field. In the gravitational field, the shared processing cores and ordinary processing cores are uniformly divided into arrays, and the lines between them form a network A lattice framework, so there may be multiple sources of gravity in the gravitational field, which attract other common processing cores to visit;

[0038]S2: Each of the common processing cores in...

Embodiment 2

[0041] On the basis of the routing mechanism for heterogeneous many-core task scheduling based on the artificial potential field method described in Embodiment 1, the step S1 specifically includes: determining the size of the gravitational source according to the parameters of the shared processing core, The law of the propagation direction of the force in the gravitational field is stipulated, and the propagation attenuation rule of the force in the gravitational field is established.

[0042] Such as Figure 4 As shown, the step S2 specifically includes: preferably, the shared processing core transmits force field data packets in the four directions of the gravitational field to the southeast, northwest, and the transmission of force is through the transmission of the force field data packets in this embodiment. To express, the normal processing core or other shared processing cores will receive the force field data packet and continue to send it, and the north, south, west,...

Embodiment 3

[0083] On the basis of the routing mechanism for heterogeneous many-core task scheduling based on the artificial potential field method described in the above embodiments, a specific embodiment is provided, such as Figure 7 As shown, there are four shared processing cores in the figure, which are used as four gravitational sources, respectively gravitational sources A, B, C, D, and their sizes are all 10. In the preferred embodiment, the "propagation of force The law of direction” is specifically: the force is first transmitted in the X dimension, and then transmitted in the Y dimension, so only in the line with the gravitational source, there is left or right force, otherwise there is only pulling force in the upper and lower directions, and the force field The establishment is first the x-axis and then the y-axis, then the direction of routing packets is the first y-axis, and then the x-axis direction.

[0084]The preferred "propagation attenuation rule of force in the grav...

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 routing mechanism for heterogeneous many-core task scheduling based on an artificial potential field. The routing mechanism comprises an S1 step of taking each shared processing core as a gravitational source to build a gravitational field; an S2 step of calculating through a potential field propagation logic unit to obtain real-time resultant forces received from the four directions of north, south, east and west in the gravitational field respectively by general processing cores; and an S3 step of comparing and routing the resultant forces received respectively by the general processing cores in the direction of the largest resultant force. According to the invention, in the task scheduling process, the routing mechanism can adjust the scheduling of request data packets of the general processing cores based on the processing loading degree change of all the shared processing cores, and conduct task scheduling first for the shared processing core with a smaller processing load to balance the task load of the shared processing cores; and in this way, reasonable scheduling and load balancing of dynamic tasks are achieved, the processing performance is improved, obstruction cases are reduced and the generation of route flapping is prevented.

Description

technical field [0001] The invention relates to a routing mechanism for heterogeneous many-core task scheduling based on an artificial potential field method, and belongs to the field of information technology. Background technique [0002] With the transition from multi-core technology to many-core technology, not only the computing nodes are gradually increasing (hundreds or thousands), but the complexity of the processor is also increasing. In order to continuously improve the performance of processor chips without breaking the relevant provisions of Moore's Law, semiconductor developers put more and more processing units on the same processor chip. Among them, including the kernel and various shared resources. Therefore, the research on heterogeneous many-core structure will become the focus in the future. In this context, it is of great significance to analyze the task scheduling problem under the heterogeneous many-core structure. In the heterogeneous many-core struc...

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): H04L12/803H04L12/867H04L47/629
Inventor 姚涛谢延华王曦爽张克松
Owner SUZHOU POWERCORE TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products