Parallel application performance vulnerability analysis method and system based on fuzzy rule reasoning

An analysis method and technology of application performance, applied in special data processing applications, electrical digital data processing, instruments, etc., can solve the problems that the analysis results are not scientific, objective, affect performance, and misjudgment rate, etc., and achieve scalability. Parallel program performance analysis, the effect of promoting refining and summarizing

Active Publication Date: 2011-12-21
THE PLA INFORMATION ENG UNIV +1
View PDF1 Cites 12 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Especially when scaling communication-intensive applications over a large number of processes, such wait states can severely impact performance
[0003] At present, statistical methods are mostly used in the analysis of parallel application performance, but there are several problems in traditional statistical methods: (1) Traditional statistical work is generally carried out by technicians based on experience and by analyzing monitoring data such as runnin

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
  • Parallel application performance vulnerability analysis method and system based on fuzzy rule reasoning
  • Parallel application performance vulnerability analysis method and system based on fuzzy rule reasoning
  • Parallel application performance vulnerability analysis method and system based on fuzzy rule reasoning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0022] Embodiment 1: The present invention’s parallel application performance vulnerability analysis method based on fuzzy rule reasoning firstly abstracts various situations that lead to inefficient parallel program performance, obtains a parallel program performance characteristic model, and constructs a parallel program performance vulnerability characteristic model reasoning rule base, and then

[0023] 1) Use the replay-based parallel trace analysis method to extract trace file features. Through distributed storage and parallel processing technology, replay the original communication on the same number of CPUs as the actual execution of the application program, so that each process can independently analyze the local Trace data, when traversing the trace data, according to the importance of each event area and related system resources, identify the required features and generate feature files;

[0024] 2) Introduce the fuzzy mathematics method into the definition and reas...

Embodiment 2

[0025] Embodiment two: this embodiment combines Figure 1 to Figure 4 , to further illustrate the specific implementation of the method for analyzing the vulnerability of parallel application performance based on fuzzy rule reasoning in the present invention. The analytical methods include:

[0026] 1. Mode definition

[0027] In order to better describe performance vulnerabilities, various execution behaviors that affect program execution performance are abstracted to form a performance model, which establishes a basis for the definition of rules.

[0028] (1) Abstract mechanism

[0029] A compound event is a collection of events in a trace file that satisfy conditions related to a specific performance problem. Although there are different types of events, each with different properties, each event provides at least a time property and a position property that represents the flow of control that caused it to occur. This representation of program execution is called the "e...

Embodiment 3

[0087] Embodiment three: see figure 1 , this embodiment is a parallel application performance vulnerability analysis system based on fuzzy rule reasoning, which realizes auxiliary decision-making for parallel application performance vulnerability analysis and discovery, and its system composition is as follows figure 1 Shown: The system is mainly composed of inference engine, knowledge base, feature extraction module, comprehensive database and user interface module.

[0088] The knowledge base stores judgmental knowledge of expert experience used for reasoning of conclusions and problem solving, as well as various control knowledge in the process of reasoning and solving. The core of the system is knowledge, and its performance depends on the quantity, quality and representation of knowledge in the knowledge base. The knowledge representation method determines the organizational structure of the knowledge base and directly affects the work efficiency of the whole system.

...

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 parallel system designing and performance optimizing technology of a high-performance computer, in particular to a parallel application performance vulnerability analyzing method and system based on fuzzy rule reasoning. The analyzing method comprises the steps of: firstly, abstracting various conditions causing low efficiency of parallel program performance to obtain a parallel program performance characteristic mode, and constructing a parallel program performance vulnerability characteristic mode reasoning rule base; secondly, extracting trace file characteristics by adopting a reset-based parallel trace analyzing method, and recognizing required characteristics to generate a characteristic file; and finally, introducing a fuzzy mathematical method into production rule definition and reasoning, automatically searching a characteristic mode of an event trace by adopting a fuzzy method-based matching mechanism, realizing matching and finding of the characteristic mode, finding and positioning a parallel program performance vulnerability, and providing aid decision making for the work of further optimizing the performance of the parallel system of the computer. The system mainly comprises a reasoning machine module, a knowledge base, a characteristic extraction module, a synthesis database and a user interface module.

Description

technical field [0001] The invention relates to high-performance computer parallel system design and performance tuning technology, in particular to a parallel application performance vulnerability analysis method and system based on fuzzy rule reasoning. Background technique [0002] In parallel system applications, processes often need to access data provided by remote processes, so that the progress of one process depends on the progress of another process, which may cause some processes to have to wait for data from other processes, or because of Idle with no more tasks to do. In point-to-point communication, the asynchrony between the sending process and the receiving process may cause the receiving process to enter the receiving state early and wait for the sending process, or block the sending process to enter the sending state early and wait for the receiving process. Collection synchronization is similar in that it requires every participating process to reach a ce...

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): G06F17/30
Inventor 于磊徐冰侯雪梅杜祝平李志博
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
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