Method and system for realizing consistency of cached data of multi-path processor based on distributed finite directory

A multi-processor, data cache technology, applied in memory systems, transaction processing, electrical digital data processing, etc., can solve the problems of starvation and blockage of request transactions, and achieve the effect of improving reliability and scalability

Pending Publication Date: 2020-09-11
NAT UNIV OF DEFENSE TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention: Aiming at the above-mentioned problems of the prior art, a method and system for realizing the consistency of multi-processor cache data based on a distributed limited directory is provided. The problem of starvation of request transactions, while providing strict execution order guarantees, ensures that in a distributed environment, data dependencies are not destroyed, data dependencies between request transactions can be completely maintained, and request transactions sent by multiple cores can be fair Coordinated execution of many-core processors improves reliability and scalability

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
  • Method and system for realizing consistency of cached data of multi-path processor based on distributed finite directory
  • Method and system for realizing consistency of cached data of multi-path processor based on distributed finite directory
  • Method and system for realizing consistency of cached data of multi-path processor based on distributed finite directory

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0031] like image 3 As shown, the implementation steps of the method for implementing the consistency of multi-processor cache data based on a distributed limited directory in this embodiment include:

[0032] 1) The private cache PCache X responds to the request of the corresponding processing unit PE X, and sends a data read and write consistency request transaction to the corresponding host DDCU X when the request misses;

[0033] 2) The host DDCU X responds to the data read / write consistency request transaction and detects whether there is a resource conflict. If there is no resource conflict, it will return data through other host DDCU or MCU, end and exit; otherwise, wait for the retry buffer [X] to be idle state free, and when the retry buffer [X] is in the idle state free, the data read and write consistency request transaction is discarded, and the retry response message RetryAck is sent to the private cache PCache X, and the next step is skipped;

[0034] 3) The fl...

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 method and system for realizing consistency of cached data of a multi-path processor based on a distributed finite directory. According to the invention, the private cache responds to the request of the corresponding processing unit; and when hitting does occurs, a data read-write consistency request transaction is sent to the corresponding host DDCU, if the host DDCU encounters a resource conflict, the messages are discarded, a flow control mode is entered into, and the discarded messages are retransmitted one by one by the private cache based on a message appointment principle until the flow control mode is exited and a normal line production mode is entered into after the resource conflict is removed. According to the invention, the problem of starvation of request affairs due to blockage of a working pipeline by directory self-replacement can be solved, meanwhile, strict execution sequence guarantee is provided, it is guaranteed that the data dependency relationship is not damaged and the data dependency relationship among the request transactions can be completely kept in the multi-processor distributed environment, it is guaranteed that the request transactions sent by the multiple cores can be fairly and coordinately executed, and then the reliability and expandability of the many-core processor are improved.

Description

technical field [0001] The invention relates to a multi-core processor architecture, in particular to a method and system for realizing data consistency of multi-processor caches based on a distributed limited directory, which is used to solve the problem that the directory-based cache (Cache) consistency protocol has a problem in the capacity of directory memory. When limited, directory self-replacement blocks the work pipeline and causes the request transaction to starve to death. Background technique [0002] The commonly used implementation method of many-core microprocessors is isomorphic integration, that is, integrating multiple mature general-purpose processing cores with the same structure and powerful functions. For example, Intel integrates 32 processor cores in the Skylake-EP Xeon E5 chip; AMD integrates 16 processor cores in the Ryzen Threadripper processor; and Feiteng series server chips integrate 16~64 processor cores. like figure 1 As shown, the many-core ...

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): G06F12/0815G06F12/0842G06F12/0875G06F12/0895G06F9/46G06F9/38G06F9/54
CPCG06F12/0815G06F12/0842G06F12/0875G06F12/0895G06F9/466G06F9/3838G06F9/546G06F2209/548
Inventor 冯权友张英周宏伟邓让钰杨乾明励楠曾坤王勇张见乔寓然王俊辉王永文
Owner NAT UNIV OF DEFENSE TECH
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