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

Parallel program performance prediction system based on runtime features and machine learning

A runtime feature and performance prediction technology, applied in instrumentation, error detection/correction, software testing/debugging, etc., can solve problems such as low accuracy, high overhead, and long prediction time, and achieve low prediction overhead and strong generalization capacity and the effect of reducing job waiting time

Active Publication Date: 2019-10-25
HARBIN INST OF TECH
View PDF7 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The present invention solves the problems of relatively large overhead, long prediction time and low accuracy in the parallel program performance prediction system based on machine learning

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 program performance prediction system based on runtime features and machine learning
  • Parallel program performance prediction system based on runtime features and machine learning
  • Parallel program performance prediction system based on runtime features and machine learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] to combine Figures 1 to 5 As shown, the realization of a parallel program performance prediction system based on runtime features and machine learning of the present invention is described as follows:

[0044] 1 Parallel Program Performance Prediction System

[0045] Such as figure 1 As shown, the parallel program performance prediction system is mainly divided into three parts: feature acquisition, performance modeling and performance prediction. The first part is the acquisition of program features, mainly by performing edge profiling instrumentation on small-scale parallel programs to obtain training data features. The program instrumentation in the present invention is based on the LLVM compiler architecture, and the program after the instrumentation is executed multiple times. Get the average value, obtain process number and basic block frequency, as the feature of training data, the total running time of program is as parallel program performance index of the p...

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 parallel program performance prediction system based on runtime characteristics and machine learning, and belongs to the technical field of parallel program performance prediction. The invention aims to solve the problems of high overhead, long prediction time and low accuracy of a parallel program performance prediction system based on machine learning. The system comprises the steps of the mixed instrumentation is performed on the original program; a basic block counter is reduced, then a program is deleted into a serial program without an input result, the runningtime of the program is shortened, meanwhile, the program execution process is reserved, the basic block frequency is accurately and rapidly obtained, the data is preprocessed and input into a prediction model, and finally the execution time of a large-scale parallel program is output. The model generated by the method has very strong generalization ability, can accurately predict the execution time of the large-scale parallel program, and is very low in prediction overhead.

Description

technical field [0001] The invention relates to a parallel program performance prediction system based on runtime characteristics and machine learning, and belongs to the technical field of parallel program performance prediction. Background technique [0002] With the rapid growth of the scale and complexity of high-performance computing systems, such as the number of nodes, storage, etc., the cost for users to execute parallel applications in high-performance computing systems also increases, and the execution of many parallel programs in high-performance computing systems The efficiency is relatively low, resulting in a waste of system resources, which makes the efficiency and scalability of high-performance systems and applications more and more prominent. Therefore, it is very important to predict the performance of massively parallel programs on target systems by running small-scale parallel programs before large-scale execution of parallel programs in high-performance...

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): G06F11/36
CPCG06F11/3608G06F11/3604
Inventor 张伟哲何慧王一名郝萌
Owner HARBIN INST OF TECH
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

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com