Heterogeneous sensing cluster scheduling method and device

A cluster scheduling and heterogeneous cluster technology, applied in the computer field, can solve the problems of not considering heterogeneity, accelerator performance and cost differences, etc., and achieve the effect of improving processing efficiency and excellent scheduling effect

Active Publication Date: 2021-07-16
ZHEJIANG UNIV +1
View PDF6 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] In current clusters, there are often multiple types of accelerators. The performance and cost of different accelerators vary greatly. Each type of AI accelerator has its corresponding advantages and disadvantages. However, the existing scheduling methods do not take this heterogeneity into consideration. characteristics, but mainly schedule the same type of accelerators

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
  • Heterogeneous sensing cluster scheduling method and device
  • Heterogeneous sensing cluster scheduling method and device
  • Heterogeneous sensing cluster scheduling method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0021] The solutions provided in this specification will be described below in conjunction with the accompanying drawings.

[0022] The present invention provides a pioneering solution, which can realize the scheduling of different types of AI accelerators in heterogeneous clusters. It should be noted that an AI accelerator can also be called an AI chip, an AI computing card, or a computing chip, or a big data computing chip.

[0023] figure 1 An implementation architecture diagram of a heterogeneous-aware cluster scheduling method according to an embodiment is shown. Such as figure 1 As shown, for the big data processing task submitted by the user, the throughput is estimated, and then combined with the preset cluster scheduling strategy, the current pending task including the big data processing task is heterogeneously aware. The scheduling method of the cluster scheduling strategy, for example, according to the characteristics of different chips such as GPU, TPU, CPU, FP...

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 embodiment of the invention provides a heterogeneous sensing cluster scheduling method. A heterogeneous cluster comprises various types of calculation chips. In the method, the throughput of the big data task is pre-estimated and used as an input, flexible definition of various cluster management strategies or automatic control modes is supported, and the big data task is better scheduled to heterogeneous hardware to be executed depending on perception of a heterogeneous environment, so that scheduling of the heterogeneous hardware is realized; and a heterogeneous sensing cluster scheduling whole body is formed, so that the purpose of efficiently utilizing cluster resources is achieved.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular, to a heterogeneous-aware cluster scheduling method and device. Background technique [0002] With the end of Moore's Law, AI (Artificial Intelligence) accelerators such as GPUs, TPUs, FPGAs, and other domain-specific accelerators (such as ASICs) have emerged as alternatives to general-purpose CPUs. The deployment of these AI accelerators has achieved great results, providing strong support for training the most advanced models and big data computing in many fields. [0003] In current clusters, there are often multiple types of accelerators. The performance and cost of different accelerators vary greatly. Each type of AI accelerator has its corresponding advantages and disadvantages. However, the existing scheduling methods do not take this heterogeneity into consideration. characteristics, but mainly schedules the same type of accelerators. T...

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): G06F9/48G06F9/50
CPCG06F9/4881G06F9/5066G06F2209/484
Inventor 朱海洋陈为周俊严凡钱中昊叶洲
Owner ZHEJIANG 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
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